PRECISE-AI Teaming Profiles
Thank you for showing an interest in ARPA-H’s Performance and Reliability Evaluation for Continuous Modifications and Useability of Artificial Intelligence (PRECISE-AI) program. This page is designed to help facilitate connections between prospective proposers.
PRECISE-AI anticipates that teaming will be necessary to achieve the goals of the program. Prospective performers are encouraged (but not required) to form teams with varied technical expertise to submit a proposal to the PRECISE-AI Module Announcement.
PRECISE-AI Teaming Profile Form
Please note that by publishing the public teaming profiles list, ARPA-H is not endorsing, sponsoring, or otherwise evaluating the qualifications of the individuals or organizations included here. Submissions to the teaming profiles list are reviewed and updated periodically.
Interested in learning more about the PRECISE-AI program?
Teaming Profiles List
To narrow the results in the Teaming Profiles List, please use the input below to filter results based on your search term. The list will filter as you type.
| Contact | Organization Name | Location | Description of Research Focus Area | Description of Teaming Partner | Technical Areas | |
| Jodi Wachs | Contexa LLC | contexahealth@gmail.com | Albuquerque, NM | Contexa is developing a domain-wide clinical ontology to address the challenge of structured vs. unstructured knowledge in AI tools for CDS. Beginning with feature abstraction and progressing to LLM development, it automates the extraction of surrogate ground truth labels across diverse clinical use cases. The ontology is continuously updated and revised based on the best evidence, enabling AI models to self-correct and implement updates, ensuring ongoing performance assessment and improvement. | We seek a partner with a clinical site and an AI CDS tool provider. Initial discussions have taken place between Contexa LLC and Jaxon.ai, which specializes in trustworthy AI that relies on ground truth data. This collaboration aims to address key challenges, including automating the extraction of surrogate ground truth labels and enabling AI models to self-correct and implement necessary updates, supporting continuous performance improvement. To do this work, we seek a prime contractor. | TA1: Automated Surrogate Ground Truth Label Extraction, TA2.3: AI Model Self-Correction Tools, TA4: Core Data Infrastructure |
| Siun-Chuon Mau | CACI, Inc. - FEDERAL | siun-chuon.mau@caci.com | Florham Park, NJ | AI/ML technologies, including ones relevant to TA1 and 3. | Clinical partners in possession of multi-modal EHR with ground-truth diagnoses and/or with capacity for field studies assessing clinical efficiency. | TA1: Automated Surrogate Ground Truth Label Extraction, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Hawk Wang | University of Mississippi | bwang3@olemiss.edu | Oxford, MS | Our research focuses on improving model robustness and generalization in deep learning and AI. Our recent work (ECCV'24) introduces a novel approach for human pose estimation in low-light conditions, eliminating the need for low-light ground truths. Another project (NeurIPS'24) centers on zero-shot human-object interaction (HOI) detection, adapting multimodal large language models for unseen classes. Both projects enable AI models to perform reliably across previously unseen environments. | We seek partners skilled in healthcare applications of AI, particularly in rehabilitation, patient monitoring, and diagnostic processes involving human pose, motion analysis, and human-object interaction. Expertise in integrating AI with clinical data systems and regulatory compliance in medical device production is valued. | TA1: Automated Surrogate Ground Truth Label Extraction, TA2: Degradation Detection & Self-Correction, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Ben Strasser | ASU Research Enterprise | benjamin.strasser@asure.org | SCOTTSDALE, AZ | ASURE (ASU Research Enterprise) specializes in AI development, AI evaluation, metamaterials, and wearable electronics. Foci include translating measures of performance to measures of effectiveness, establishing model-independent user- and theory- driven assessment methodologies, and developing real-time performance evaluation tools for in-situ assessment. These efforts are led by a former DARPA program manager with strong track record in AI evaluation. | Our capabilities include ensuring statistical validity of results and methodologies, validating these results with independent data, and overseeing adherence to engineering and interoperability standards. We seek partners with clinical expertise to ensure clinical relevance of results and interface with clinical sites to enable collection of independent verification and validation data. | TA5: Independent Verification and Validation |
| Yonghui Wu | University of Florida | yonghui.wu@ufl.edu | Gainesville, FL | Natural language processing, large language model, and multi-modal model to generate surrogate labels using clinical notes and medical images for various medical AI applications such as patient information extraction, clinical phenotyping, disease onset prediction, treatment plan generation. We have strong track records in study healthcare disparity using patient-level and contextual social determinants of health. | TA2 and/or TA3 | TA1: Automated Surrogate Ground Truth Label Extraction, TA2: Degradation Detection & Self-Correction |
| William Walton | JHU/APL | william.walton@jhuapl.edu | Laurel, MD | We have developed uncertainty estimation techniques for AI/ML algorithms in the medical domain. One application involves breast cancer detection. We have a multi-modality (fusion) breast cancer detection algorithm and a CNN Dual View (CC/MLO) X-ray image registration algorithm, both of which provide uncertainty estimates with their outputs. The uncertainty estimates can indicate how well the algorithms are performing. The algorithms are extendable. (See our publications online.) | We seek partnership with those focusing on TAs 1, 2, and 3. For example, our uncertainty-based algorithms and methodologies could be used in conjunction with or in support of TA-1, automated ground truth label extraction; TA-2, model degradation and self-correction, or in collaboration with other TA-3 efforts for enhancing clinician trust in AI and decision-making processes in improving overall clinical performance. Our methodologies can be extended to other domains. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Andrew Ritcheson | ICA | aritcheson@ica.ai | Arlington, VA | Proven track record in the regulated medical device environment and deep understanding of regulatory processes. Current work involves the application of advanced NLP an ML techniques to extract data from structured and unstructured documents. Actively fine-tuning advanced transformer models (e.g., LLMs) for complex, context-based insights extractions. Multidisciplinary teams with extensive experience applying agile project management to lead large, complex federal research programs. | We are looking for teaming partners with access to real-world data representing diverse populations and clinical informatics capabilities. We are also interested in those with in-house SMEs in internal medicine and specialty areas relevant to the use-cases of PRECISE AI. | TA1: Automated Surrogate Ground Truth Label Extraction |
| Eean Patterson | Judy.ai | eean@judy.ai | Phoenix, AZ | Judy Ai is an integrated Ai system that assists in clinical judgement and organizational waste. | Working models to integrate into judy, Assist design team in development of UXUI interface, Align with commercial partners, medical Organizational management | TA3: Quantify Uncertainty & Improve Clinician Performance, TA2.2: AI-Based Root-Cause-Analysis Tools, TA4: Core Data Infrastructure, TA1: Automated Surrogate Ground Truth Label Extraction |
| Neda Moghim | ODU | nmoghim@odu.edu | Suffolk, VA | We work on applying ML algorithms in various domains and their performance accuracy and stability. | We look for experts working on healthcare and bioinformatic fields. | TA2: Degradation Detection & Self-Correction |
| Shirui Luo | Center for AI Innovation at NCSA at the University of Illinois | shirui@illinois.edu | 1205 W Clark St, Urbana, IL 61801, IL | Our team specializes in generative AI for dense 3D medical image distributions, learning complex multimodal images with 3D consistency across tasks like modality translation, super-resolution, segmentation, and anomaly detection. These methods also support domain shift detection and adaptation. Additionally, we have expertise in hosting large language models (LLMs) and using retrieval-augmented generation (RAG) for domain-specific chatbot applications. | We are looking for partners with clinical domain expertise who can help identify real-world applications where addressing model degradation and self-correction is essential. | TA2: Degradation Detection & Self-Correction, TA2.3: AI Model Self-Correction Tools, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Thien Nguyen | University of Oregon | thienn@uoregon.edu | Eugene, OR | We design effective learning algorithms for Information Extraction, Text Mining, Large Language Models, Representation Learning, Active Learning, Natural Language Processing, and Data Mining, with applications spanning diverse domains, modalities, and languages, including clinical and biomedical NLP. | AI researchers and clinicians to a winning team for TA1, TA2, and T3 | TA1: Automated Surrogate Ground Truth Label Extraction, TA2: Degradation Detection & Self-Correction |
| Fleming Lure | Caddie Technology Inc (https://caddieai.io/) | fleming.lure@caddieai.io | Potomac, MD | We developed Smart Imagery Framing and Truthing (SIFT) system to provide automatic, rapid, fine-grained labelling and annotation of 69 and 47 abnormalities on CXR and chest CT, respectively. We have used SIFT to annotate over 450,000 publicly available CXR and chest CT collected by MIDRC, TB Portals/NIAID, Clinical Center of NIH, Stanford U, NLST, etc. We provide accurate, fast, and affordable annotation service for ML/AI researchers to train AI and clinical study for FDA clearance. | Caddie is seeking partners in TA1, TA4, and TA5, especially in terms of annotation and clinical expertise. | TA1: Automated Surrogate Ground Truth Label Extraction, TA5: Independent Verification and Validation, TA4: Core Data Infrastructure |
| Aysegul Ucar | Firat University | agulucar@firat.edu.tr | Elazig, Turkey | Our research focuses on autonomous robotics, AI, and computer vision, with applications in GPS-denied environments, multi-modal SLAM, and 3D mapping. Key projects target disaster response, agriculture, and healthcare, utilizing transformer-based object recognition and multi-sensor data fusion. We prioritize knowledge transfer and align research with industry needs through partnerships and standardization efforts, aiming for impactful, real-world solutions. | We seek teaming partners with expertise in AI, robotics, and healthcare who excel in real-world applications, such as multi-sensor fusion and autonomous systems. Ideal collaborators are innovative, industry-aligned, and committed to knowledge transfer, enhancing our focus on impactful solutions for complex environments, including disaster response and remote healthcare. We value partners who bring complementary skills to drive our research toward practical and scalable applications. | TA2.2: AI-Based Root-Cause-Analysis Tools, TA2: Degradation Detection & Self-Correction, TA5: Independent Verification and Validation |
| Carlos Goncalves | University of Colorado-Anschutz Medical Campus | carlos.goncalves@cuanschutz.edu | Aurora, CO | The Transplant Department at UCHealth is developing AI models to predict graft quality from deceased cardiac donors and to assess graft survival outcomes after transplantation. | Ideal partners will have a strong track record in healthcare, data science, or AI, with experience in organ transplantation or similar clinical fields. | TA3: Quantify Uncertainty & Improve Clinician Performance, TA2.3: AI Model Self-Correction Tools |
| Mwisa Chisunka | John Snow Labs | mwisa@johnsnowlabs.com | Lewes, DE | John Snow Labs provides state-of-the art software, models and data to help healthcare and life science organization put AI to good use. The team develops the open-source Spark NLP library (40,000 + models, 100 + millions downloads, 250+ languages) and LangTest (100+ test types for Responsible AI) and commercially the Healthcare NLP library, Healthcare-GPT LLM, the Generative AI Lab No Code Platform, and Medical Chatbot. | For TA2, a university or research institution with a strong track record in "traditional" machine learning, biostatistics, and clinical research. For TA4, a technical partner with regulated data platform expertise, particularly in asset versioning. We are also seeking partners with publication interest and partners with experience in patent support or intellectual property development in clinical or AI fields. | TA1: Automated Surrogate Ground Truth Label Extraction, TA3: Quantify Uncertainty & Improve Clinician Performance, TA5: Independent Verification and Validation, TA4: Core Data Infrastructure |
| Imanuel Lerman | InflammaSense, UCSD | ilerman@inflammaSense.com | La Jolla, CA | Our multidisciplinary teams are focused on developing GAPS-WARS (Global Autonomous Pathogen Sentinel Widely Augmented with Replicants and Simulants) to transform pathogen detection and response across diverse hospital environments. Using a federated AI framework and integrating synthetic pathogen-host simulants, GAPS wars aims to provide real-time, privacy preserving diagnostics and clinician decision support. | Our team is looking for partners for TA3-4. | TA1: Automated Surrogate Ground Truth Label Extraction, TA2: Degradation Detection & Self-Correction |
| Micah Guffey | Aneurisk Inc. | guffey@aneurisk.ai | Pittsburgh, PA | AI/ML-based predictive modeling of disease progression | Partners with access to clinical imaging and correlated outcome data for arterial aneurysm disease. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Ben Kogan | Arctop | bk+1@arctop.com | Los Angeles, CA | Arctop specializes in real-time brain decoding technology, enabling applications that interpret human cognition and emotions. Our research focuses on developing adaptive systems that respond to cognitive states, enhancing human-computer interactions. We aim to integrate our expertise in brain-computer interfaces with AI model monitoring to detect and mitigate performance degradation, aligning with ARPA-H's PRECISE-AI objectives. | Arctop seeks partners with expertise in healthcare data integration, advanced AI model monitoring, and clinical validation. Ideal collaborators will have experience in interdisciplinary projects in healthcare, neuroscience, and regulatory compliance, to co-develop robust, adaptive AI solutions. We're looking for teams who share our commitment to advancing noninvasive brain-computer interfaces and precision health. | TA3: Quantify Uncertainty & Improve Clinician Performance, TA2.2: AI-Based Root-Cause-Analysis Tools |
| Hongfang Liu | University of Texas Health Science Center at Houston | hongfang.liu@uth.tmc.edu | Houston, TX | Open Health AI Assembly, previously, Open Health Natural Language Processing Consortium focuses on the federated development, evaluation, and deployment of data extraction and curation in establishing ground truth. We establish the TIES-RITE-FAIR (teamwork, innovation, excellence, stewardship for team science, reproducible, implementable, transparent, explainable for translational science, and findable, accessible, interoperable, reusable for data science) framework in healthcareAI evaluation. | We look for potential teaming partners who align with our translational AI framework. | TA1: Automated Surrogate Ground Truth Label Extraction, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Erin Mueller | Medical Imaging and Data Resource Center (MIDRC) | erin.mueller@bsd.uchicago.edu | Chicago, IL | MIDRC (midrc.org) is an imaging data commons that has ingested 500K+ imaging studies from institutions around the nation. MIDRC has curated and published 177K imaging studies on data.midrc.org for free and open use, along with various algorithms, tools and resources. MIDRC aims to foster machine learning innovation through its ecosystem of data sharing via rapid and flexible collection, curation, harmonization, analysis and dissemination of representative imaging and associated clinical data. | While MIDRC currently has 100+ investigators at 20+ institutions with much expertise, we are always open to discussion and collaborating with others with complementary expertise. Collaborators with imaging studies are welcome to contribute cases to grow MIDRC across a variety of modalities and diseases/conditions. In addition, use cases for furthering our sequestered data and validation/translational/regulatory processes, as well as our various open tools/algorithms, are also needed. | TA5: Independent Verification and Validation, TA1: Automated Surrogate Ground Truth Label Extraction, TA3: Quantify Uncertainty & Improve Clinician Performance, TA4: Core Data Infrastructure |
| Ravi Madduri | University of Chicago and Argonne National Laboratory | madduri@uchicago.edu | Chicago, IL | The UChicago-Argonne team has been at the forefront of developing foundational building blocks of applying AI to challenges in Biomedicine. Supported by multi-year, multiple millions of dollar support from Department of Energy, Department of Veterans Affairs and the National Cancer Institute, the team has built a legacy of expertise in secure data management, data wrangling, model building, and the entire lifecycle of AI—from validation to deployment. | We are looking for consortiums, healthcare systems that can provide a collaborative environment and commitment to apply multimodal foundational AI models to build a true learning health care system. | TA2: Degradation Detection & Self-Correction, TA5: Independent Verification and Validation, TA3: Quantify Uncertainty & Improve Clinician Performance, TA4: Core Data Infrastructure |
| Asad Mansoor | Cloud Control Studio Inc | asad@cloudcontrolstudio.com | Washington DC, Fredrick Maryland, and the rest of DC metro area., VA | At Cloud Control Studio, we’ve developed a AI- Risk Evaluation Framework with over 100 controls to assess and quantify both technical and non-technical risks in AI-based technologies, addressing a gap in industry auditing. Our controls align with NIST AI RMF guidelines and industry best practices, providing a comprehensive framework that ensures AI tools are fully auditable, mitigating risks and enhancing compliance in AI-based products within the public sector and healthcare industries. | We are looking for teaming partners that will support majority of the PRECISE-AI work, however need support regarding security, risk and compliance management for AI usage to ensure secure, responsible and safe implementation of all capabilities AI has to offer. | TA3: Quantify Uncertainty & Improve Clinician Performance, TA2.3: AI Model Self-Correction Tools, TA2.2: AI-Based Root-Cause-Analysis Tools, TA5: Independent Verification and Validation |
| Allison Dempsey | University of Colorado School of Medicine | allison.dempsey@ucdenver.edu | Denver, CO | Our team specializes in (1) clinical application of computation models and employing linear and nonlinear ensemble modeling, which is model naiive, but corrects component models by their error; (2) measurement theory with experience in developing novel mathematical representations and algorithms, as well as benchmarking reliability translated across technical fields; and (3) clinical measurement development/collection, technical integration, and cross-institution data conditioning and use. | We are seeking to team with partners with expertise in TA1 (to generate input into our evaluation tool) and TA 4/5 (to assess implementation). | TA3: Quantify Uncertainty & Improve Clinician Performance, TA2.1: Continuous Degradation Detection Tools, TA4: Core Data Infrastructure |
| Arkady Hagopian | Syntropy Technologies, LLC | arkady.hagopian@emdgroup.com | Cambridge, MA | Syntropy Technologies focuses on healthcare data interoperability, secure data sharing, and advanced analytics. Our primary research areas include deploying scalable data infrastructure for healthcare ecosystems and advancing federated data solutions for clinical and biomedical research. | We are interested in speaking with individuals and organizations who can navigate complex decision support regulatory landscapes, administer and manage large-scale data projects (including budgets), and align with our mission to deliver secure, interoperable data solutions for clinical environments. | TA4: Core Data Infrastructure |
| Nicholas Bedworth | SocialEyes Corporation | nicholas.bedworth@socialeyesus.com | Boston, MA | Scalable AI-driven global healthcare platform for low-income populations, including hand-held, non-invasive, multi-disease point-of-care diagnostic devices, apps, infrastructure. Digital twins are used to model human tissue anatomy and physiology, photonics-based acquisition devices, and synthetic datasets. The entire solution is designed with inherent AI-governance as well as adaptibility in mind, along with the ability to operate at massive scale across diverse patient populations. | MD/PhD experts in modeling tissues using volumetric datasets; wave optic light transport rendering; regulatory strategy; digital twin scientists, technologists and experimentalists; AI computational imaging. | TA1: Automated Surrogate Ground Truth Label Extraction, TA3: Quantify Uncertainty & Improve Clinician Performance, TA5: Independent Verification and Validation |
| Jason Kramer | ObjectSecurity | jason@objectsecurity.com | San Diego, CA | ObjectSecurity’s research, partially funded by a Phase II Air Force SBIR, focuses on AI/ML explainability and security, ensuring trustworthiness by assessing risks and evaluating performance. We address security threats and compliance challenges in AI systems through automated assessments that analyze behavior, detect vulnerabilities, and recommend improvements for robustness and accuracy, while ensuring regulatory compliance. | We plan to address the AI model self-correction tools to predict performance risks (TA2.1, 2.3), recommend optimal updates, and ensure safe, automated model adjustments while maintaining compliance and patient safety. We seek teaming partners with expertise in continuous degradation detection, root cause analysis, and safeguards for automated updates to help build a comprehensive AI monitoring and correction pipeline. | TA2.3: AI Model Self-Correction Tools, TA2.1: Continuous Degradation Detection Tools |
| Shirali Nigam | Cogitamentum Bio | shirali@cogitamentumbio.com | Vienna, VA | At Cogitamentum Bio, we combine rigorous research and medical expertise with strategic business acumen to drive the successful commercialization of biotechnologies. We specialize in integrating our robust healthcare experience to support high-impact applications of novel technologies in this sector, and position them well for successful commercial deployment and growth. | Our team brings extensive experience from prior roles in 'Access to and Quality of Care', having collaborated with state governments, insurances, patient groups, providers, CMS, and government officials. Their medical experience complemented by work on rollouts for COVID-testing and vaccines, positions our team well to identify clinically relevant subpopulations, uncover root-cause degradations, and recommend actionable mitigation strategies for the implementation of TA2. | TA2: Degradation Detection & Self-Correction, TA1: Automated Surrogate Ground Truth Label Extraction |
| Mike McMahan | NGZ AI, Inc. | mmcmahan@ngz.ai | Washington DC & Cambridge, MA, DC | Creating AI models tailored for the Department of Defense and Veterans Affairs, supported by research teams from Cambridge, MA, specializing in radiology, CT scans, and traumatic brain injury (TBI). | NGZ is looking to potentially become part of a larger team. Our solutions are focused on all 3 aspects of TA2. | TA2: Degradation Detection & Self-Correction, TA2.1: Continuous Degradation Detection Tools, TA2.2: AI-Based Root-Cause-Analysis Tools, TA2.3: AI Model Self-Correction Tools |
| Valerio Pascucci | University of Utah | pascucci.valerio@gmail.com | Salt lake city, UT | Data sharing and cyberinfrastructure were developed with an emphasis on a federated model allowing remote use and different levels of security. | Looking for other institutions with expertise in specific areas that may be part of a national federated model. (1) images (2) medical datyabses, (3) identification, (4) federated learning, ... | TA4: Core Data Infrastructure |
| Selen Bozkurt | Emory University | selen.bozkurt@emory.edu | Atlanta, GA | As a health informatics researcher, my lab specializes in real-world data (RWD), information extraction from clinical narratives, and evidence generation to improve care quality and promote health equity. We work with EHRs, registries, and RCT data to develop and test predictive models. Additionally, we contribute ML governance pipelines to ensure responsible AI use within EHRs. | small businesses with complementary expertise, particularly in developing and testing AI solutions. Ideal partners would bring strong methodologies to refine and implement solutions using EHRs. | TA1: Automated Surrogate Ground Truth Label Extraction, TA2.2: AI-Based Root-Cause-Analysis Tools, TA2: Degradation Detection & Self-Correction |
| Anjum Khurshid | Harvard Pilgrim Health Care Institute | anjum_khurshid@hphci.harvard.edu | Boston, MA | AI and biomedical informatics research for population health research and public health surveillance using electronic medical records and social determinants of health data in real world settings. | AI and LLM scientists with deep knowledge of AI model characteristics, testing, and implementation. | TA4: Core Data Infrastructure, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Sharon Ricks | Heudia Health | sharon.ricks@heudia.com | East Stroundsburg, PA | Heudia delivers a subscription-based suite of health care navigation and social support tools that transform how people get and stay healthy. We offer the health ecosystem a swift, confident connection to health care providers, the community health workforce, and social services. We utilize a hybrid recommender with a content-based foundation as well as AI and NLP techniques for feature extraction to rank healthcare services based on a person’s assessment results. | We are looking for teaming partners who are interested in improving the ability of their health system to connect patients to social and health services utilizing our AccessMeCare Insights tools proven to reduce avoidable costs of care by 20%. Partners who are committed to improving quality of care, reducing costs, and reducing health disparities by addressing access to clinical care, health behaviors, or the social drivers of health are our top priority. | TA2.2: AI-Based Root-Cause-Analysis Tools, TA1: Automated Surrogate Ground Truth Label Extraction, TA4: Core Data Infrastructure |
| Christopher Carr | RSNA | ccarr@rsna.org | Oak Brook, IL | Multimodal data collection, de-identification and labeling for AI research focusing on medical imaging. Tooling for imaging data de-identification and processing Extraction of labels from multimodal data sources including through use of foundational models. Incubation of model development through use case-specific challenge competitions. Validation and statistical analysis of model performance, including bias and generalizability assessment and comparison of AI and human performers. | Tooling and expertise for extraction and management of non-imaging clinical variables. Expertise in root cause analysis of model degradation and self-correction methods. Expertise in data gathering for clinician performance. | TA1: Automated Surrogate Ground Truth Label Extraction, TA2: Degradation Detection & Self-Correction |
| Armend Cobovic | Dandelion Health | armend@dandelionhealth.ai | New York, NY | As a potential TA4 as well as a data contributor for TA1 Dandelion’s multimodal Real World Data Platform brings together clinical notes, images and waveforms from millions of patient records for the most comprehensive view of disease progression, treatment efficacy, and individual patient outcomes allowing users to easily develop AI solutions. | Looking for partners that could contribute to the TA2 and TA3 portions of the program. | TA4: Core Data Infrastructure, TA1: Automated Surrogate Ground Truth Label Extraction |
| Nitin Jain | InnoVet Health LLC | nitin@innovethealth.com | Dallas, TX | Innovet Health (SDVOSB) conducts IV&V services as a Prime contractor at VA for target error and risk factor detection in federal IT systems such as financial applications. As a prime contractor to VA we also bring AI/ML expertise in analyzing health, operational & financial data to identify patterns of health outcome risks and operational and financial efficiencies. We combine the power of AI/ML and IV&V to automate AI/ML models validation without human intervention. | Looking for partners experienced in federal policy, risk management and AI/ML modeling | TA5: Independent Verification and Validation, TA2: Degradation Detection & Self-Correction |
| David Dorr | OHSU | dorrd@ohsu.edu | Portland, OR | OHSU, as the sole Academic Health Center in Oregon, has a strong history of informatics, data science, and machine learning innovation, evaluation, and implementation. We have an advanced AI governance and support for a living laboratory for AI models, with a goal to implement and benchmark advanced approaches for quality assurance, monitoring, and self-correction and evaluate results. | Those who develop the self-correcting and benchmarking approaches, as well as model developers. | TA4: Core Data Infrastructure |
| Wei Xu | Georgia Institute of Technology | wei.xu@cc.gatech.edu | Atlanta, GA | Expertise in Natural Language Processing research, focusing on (1) analyzing and evaluating large language models; (2) information extraction and domain adaptation. | A good partnership with complementary expertise. | TA2: Degradation Detection & Self-Correction |
| Mohammad Adibuzzaman | Oregon Health and Science University | adibuzza@ohsu.edu | Portland, OR | Core data infrastructure for real time AI governance in the health system with a focus on ethical aritficial intelligence for healthcare. We are also working to be an organization for independent organization for benchmarking and validation for other research organizations and commercial companies. We are strong with EHR, data analysis, implementation science and data governance as an academic medical center. | Technical expertise on root cause analysis and degradation monitoring. | TA5: Independent Verification and Validation |
| Osama Tarraf | istek LLC | otarraf@istekllc.com | Fairfax, VA | Our research focuses on AI-driven solutions, including predictive analytics, smart automation, and image recognition. We develop automated tools for real-time performance monitoring and degradation detection in AI systems. Our expertise in machine learning, data extraction, and AI based self organizing and optimizing algorithms directly aligns with continuous monitoring and self-corrective mechanisms. | We are seeking highly skilled partners with expertise in healthcare data systems, clinical validation, and regulatory compliance to complement our leadership in machine learning, data extraction, and AI-based self-optimizing algorithms. Together, we will drive success in TA1 and TA2, creating cutting-edge solutions. | TA2: Degradation Detection & Self-Correction |
| Stuart Milner | SMILNERS, Inc. | stu.milner@gmail.com | Potomac, MD | AI | AI | TA2: Degradation Detection & Self-Correction |
| Tajuddin Manhar Mohammed | Mayachitra, Inc. | mohammed@mayachitra.com | Santa Barbara, CA | Mayachitra specializes in advanced image processing, computer vision, and AI for content-centric applications. We focus on visual content search, image classification, and robust ML model development. Our emphasis on performance monitoring, degradation detection, and adaptive learning aligns with PRECISE-AI's goals for maintaining AI model accuracy in clinical settings. | We seek partners with expertise in healthcare informatics, clinical AI, and medical imaging. Ideal collaborators would have experience in establishing clinical "ground truth" datasets, quantifying model uncertainty, and communicating AI insights to healthcare professionals. We're interested in partners with access to diverse clinical environments for real-world AI testing. | TA2.3: AI Model Self-Correction Tools |
| Jay Pujara | University of Southern California | jpujara@usc.edu | Los Angeles, CA | AI, knowledge graphs, contextual learning, forecasting, data integration, conversational AI, persona-based modeling | medical expertise | TA2.3: AI Model Self-Correction Tools |
| Scott Cohen | Jaxon, Inc. | scott@jaxon.ai | Boston, MA | Jaxon is a set of cooperative technologies that provide robust verification and validation guardrails for genAI outputs. It enables formal reasoning and proof-based validation of AI responses, minimizing risks associated with hallucinations or incorrect outputs. DSAIL’s modular runtime spectrum includes Retrieval Augmented Generation (RAG)-based fact-checking, Bayesian belief networks for evidence-based reasoning, and symbolic reasoning for rigorous logical proofs. | We would like to be part of a team | TA2.3: AI Model Self-Correction Tools |
| Bijan Tadayon | Z Advanced Computing Inc | bijantadayon@zadvancedcomputing.com | Potomac, MD | Since 2011, we have been working on a new AI / ML set of algorithms for 3D object/image recognition, using typically only 10 to 50 training samples (instead of 1000s to billions required by Neural Nets used by other companies), with better accuracy & much smaller hardware/ CPU/ GPU, already demonstrated for Aerial Images for US Air Force & Smart Appliances for Bosch. To our knowledge, we are the only company in the world that can train AI using only a few training samples. | World-renowned team of scientists, e.g., Prof Lee, Nobel Prize in Physics & late Prof Zadeh, AI-Hall-of-Fame & Prof Peyman (Inventor of LASIK - awarded a prize by US President). Have over 450 inventions, including 14 issued US patents. Have done 14 invited tech lectures at the US Patent Office on AI/ ML subject (more than any other researcher). Looking for pathologists & other physicians to team up for medical imaging for cancer & many other medical areas. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Jim StClair | MyLigo, Inc | jim.stclair@myligo.io | Austin, TX | Research focus includes enabling federated AI, Digital Twins, and local AI/ML integration | Seeking other technical partners and government contractors to position for development opportunities | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Mark Kahan | Synopsys, Inc. | mkahan@synopsys.com | Marlborough, MA | We are developing a number of novel ophthalmic methods that could dramatically improve disease detection/identification that employ low HW SWaP. | Manufacturers of Novel Solid-State Ophthalmic devices covering broad spectral ranges and digital twins. | TA1: Automated Surrogate Ground Truth Label Extraction |
| Marco Smit | Gesund.ai | marco@gesund.ai | Palo Alto, CA | Gesund.ai's platform eliminates the trade-off between regulatory compliance/rigor an speed & flexibility of medical AI development. Our current research focus is on a) massively speeding up the emergence of AI Assurance Labs b) expand use of our platform for post-market monitoring and validation, especially for providers with diverse populations. Gesund.ai wants to democratize fair, equitable access to high-quality AI for all providers | We would be looking for partners who have relevant data, those with regulatory expertise, but also providers and AI developers who want to use this initiative to be better prepared for the next 'new normal' with ubiquitous, responsible, and agile use of AI in the clinic. | TA2: Degradation Detection & Self-Correction |
| Greg Sorensen | RadNet | sorensen@radnet.com | Somerville, MA, but with practices at 400 locations in 8 states across the USA, MA | Using deep learning to bring the best care possible to each patient. We have substantial areas of AI research in population screening including cancer (breast, lung, prostate) and other disease. | Expertise in advanced deep learning algorithms and science for medical imaging, especially epidemiological applications of deep learning in imaging. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Christopher Idelson | ClearCam Inc | cidelson@clearcam-med.com | Austin, TX | Minimally Invasive Surgery Visualization | Partners with extensive data familiarity with surgical video as well as ML expertise. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Jingtong Hu | University of Pittsburgh | jthu@pitt.edu | Pittsburgh, PA | On-Device Learning, Medical Applications | On-Device Learning, Medical Applications | TA2: Degradation Detection & Self-Correction |
| Armend Cobovic | Dandelion Health | armend@dandelionhealth.ai | New York, NY | Dandelion Health’s multimodal Real World Data Platform brings together disparate data from across all care settings to build the most complete picture of patient outcomes over time. Go beyond traditional real world data, such as claims and ICD codes, and access the full range of data modalities, including structured EHR data, imaging and unstructured data types. | Looking for partner's that will expedite the vision put forth in the PRECISE-AI program. | TA1: Automated Surrogate Ground Truth Label Extraction |
| Mark Zarella | Mayo Clinic | zarella.mark@mayo.edu | Rochester, MN | Model evaluation, AI risk, Clinical explainability, technical validation, robustness measurement, human-AI interaction | Synchrony and consensus in establishing best practices in this area, academic collaboration (eg trainees) | TA5: Independent Verification and Validation |
| Ismini Lourentzou | University of Illinois Urbana - Champaign | lourent2@illinois.edu | Champaign, IL | Our research focuses on multimodal machine learning (natural language processing and computer vision), particularly for healthcare applications. Our lab has designed new medical imaging AI methods for localized disease detection and progression, robust self-supervised contrastive learning algorithms, and chest X-ray (CXR) interpretation AI models guided by the radiologists' attention patterns through eye-tracking. We have expertise to assist teams in TA1, TA2, and TA3. | Seeking teaming partners with expertise in complementary areas such as clinical validation and healthcare operations. | TA2: Degradation Detection & Self-Correction |
| Nathan Hotaling | Axle Informatics | nathan.hotaling@axleinfo.com | Rockville, MD | Creating scalable web apps focused on visualization & human-in-the-loop validation on high dimensional datasets (structured/unstructured). Building comprehensive multimodal clinical and pre-clinical biomedical datasets that can be linked using privacy preserving methods. Real world evidence QA/QC, harmonization, causal analysis, and AI/ML powered classification/cohort ID/validation/etc. Leveraging LLMs on unstructured data for element extraction, summarization, knowledge graph creation, etc. | Data providers, linkage experts, secure flexible data analysis platforms, ground truth clinical annotations of unstructured data (images, clinician notes, procedure notes, etc.). | TA4: Core Data Infrastructure |
| Thomas Grist | University of Wisconsin | thomas.grist@wisc.edu | Madison, WI | Improving human health through development and application of AI in medical imaging | Model development, validation, and scaling, quantifying uncertainty, ongoing quality control to assess drift. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Lifu Huang | University of California, Davis | lfuhuang@ucdavis.edu | UC Davis, CA | Natural Language Processing, Multimodal Learning, Explainable and Trustworthy AI, Uncertainty Quantification | Clinicians | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Steve Pieper | Isomics, Inc. | pieper@isomics.com | Cambridge, MA | Application of computing technologies generally and AI specifically to clinical problems involving imaging and therapy. | Project management and government contracting for scalable research and transformative development efforts. | TA1: Automated Surrogate Ground Truth Label Extraction, TA4: Core Data Infrastructure, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Beatrice Knudsen | University of Utah | Beatrice.Knudsen@path.utah.edu | Salt Lake City, UT | Research area: computational pathology, algorithm development, automated pixel-level cell annotations, classification, segmentation, pathology slide restauration, virtual multiplexed immunohistochemistry, evaluation metrics development | We are interested in collaborators who are deploying clinical grade algorithms for CT | TA1: Automated Surrogate Ground Truth Label Extraction, TA2.1: Continuous Degradation Detection Tools, TA3: Quantify Uncertainty & Improve Clinician Performance, TA4: Core Data Infrastructure |
| Baxter Eaves | Redpoll | bax@redpoll.ai | St Louis (and remote), MO | We develop scalable and interpretable Bayesian AI/ML methodologies for deployment alongside human actors in high risk domains. We focus on safe, online models that (properly) quantify uncertainty and handle sparse and heterogenous data. We have performed on DARPA SAIL-ON (detect and correct model degradation from drift), ECoSystemic (detect and attribute high-dimensional data inconsistency), and others. | Health domain expertise and data. | TA2: Degradation Detection & Self-Correction, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Matthew Molenda | AM Operating LLC | molenda@anatomymapper.com | Holland, OH | We developed and maintain a source of truth for AI, generating research-ready and interoperable health data through simplified inputs. Accurate, precise, and reproducible anatomic site data is a key missing variable in current standards needed for tracking of cancer data over time. Anatomy Mapper models support over 100 languages and our base language has been adopted into ICD-11. AM models also automatically enhance the language with simplified inputs, facilitating tracking of diseases. | A development partner to help move our models forward. We already have a custom model under development related to TA 1 that focuses on the automatic extraction and integration of data across different clinical use cases to establish a “ground truth” about each patient. We are a small business and SBIR recipient. | TA1: Automated Surrogate Ground Truth Label Extraction, TA4: Core Data Infrastructure |
| Andres Rojas | Azurix Advisors | Info@azurixadvisors.com | n/a, MD | Azurix Advisors, a Veteran and Minority-Owned business, brings deep expertise in healthcare strategy, technology implementation, and data analytics, making us a valuable partner to support goals of the Precise-AI program. With our extensive experience consulting across the healthcare spectrum - working with payers, providers, life sciences, and healthcare technology enablers- we are uniquely positioned to contribute to the successful development and execution of AI-driven solutions. | We are eager to explore teaming opportunities with other organizations attending PRECISEAI. We are prepared to collaborate as a subcontractor or joint partner, offering our expertise in program management, strategic development, and AI Implementation to help ARPA-H achieve its program milestones. Our cross-functional team of experts are capable of addressing technical challenges while driving innovations in AI usability and performance. | TA4: Core Data Infrastructure |
| HAN-PANG CHIU | SRI International | han-pang.chiu@sri.com | Princeton, NJ | AI/ML technologies | clinical partners | TA2: Degradation Detection & Self-Correction |
| James Buggy | VADUM Inc | james.buggy@vaduminc.com | Raleigh, NC | Vadum's expertise is bringing concepts to life in the form of products or prototypes to address challenging national defense problems. We have leveraged our experience in the Defense arena to address AI/ML solutions in the health field. | Teaming partners would leverage our experience in AI/ML to solve challenging problems in the health industry. | TA2.2: AI-Based Root-Cause-Analysis Tools |
| Kun Yang | SMU | kunyang@smu.edu | SMU, Dallas, TX 75205, US, TX | Our current research focuses on applying machine learning and artificial intelligence to solve real-world challenges. Key areas include CV, LLM and federated learning, with an emphasis on healthcare applications, data privacy, and model optimization. We are also exploring AI-driven solutions for sustainability, human-computer interaction, and enhancing decision-making processes across industries. | We seeks partners with expertise in AI, machine learning, and data science who share a commitment to innovation and real-world impact. | TA2.2: AI-Based Root-Cause-Analysis Tools |
| Michael Brown | Trail of Bits | michael.brown@trailofbits.com | Remote (US), NY | Our team consists of leading-edge AI/ML model and systems security experts. Model robustness is a focus area for us - we have developed novel capability evaluation frameworks for AI/ML systems in prior work for sponsors such as the UK's AISI and NCSC. | We are looking to support a team that is well positioned within the healthcare sector / NIH who is looking for SMEs on AI/ML system capability evaluation / robustness measurement. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Todd McCollough | Ellumen Inc. | tmccollough@ellumen.com | Silver Spring, MD | Data sharing platform, deidentification pipeline for medical images and text, make data available to interested parties | Hospitals, government medical facilities, sharable data, TA3: Quantify Uncertainty & Improvement of Clinician Performance | TA4: Core Data Infrastructure |
| Haixu Tang | Indiana University Bloomington | hatang@iu.edu | Bloomington, IN | We work on the development of machine learning and AI algorithms for biomedical research. are interested in the safety and privacy issues in AI systems. | Domain experts in biomedical AI applications | TA2.2: AI-Based Root-Cause-Analysis Tools |
| Vida Abedi | Penn State University | vabedi@pennstatehealth.psu.edu | Hershey, PA | My lab is focused on Artificial intelligence for Health Care applications with a focus on Health Disparity and Improving Data Quality for AI/ML Readiness. | Collaborators on building more equitable solutions to address the many health care challenges affecting minorities and under-represented communities. | TA1: Automated Surrogate Ground Truth Label Extraction |
| Yuankai Huo | Vanderbilt University | yuankai.huo@vanderbilt.edu | Nashville, TN | Our team focuses on the development of AI to improve imaging configuration and holistic hardware and software design. The high-field 3T MRI and ultra-high-field 7T lead offer a greater signal-to-noise ratio but also come with challenges such as increased artifacts, susceptibility effects, and higher costs. AI can be leveraged to detect the degradation and self-correct such artifacts with minimal manual efforts, maximizing clinical performance. | We are looking for potential teaming partners who are interested in AI for optimal imaging with holistic hardware and software design, especially for 3T and 7T MRI. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Shiming Yang | University of Maryland School of Medicine | syang@som.umaryland.edu | Baltimore, MD | The Shock Trauma Research Lab in University of Maryland Baltimore maintains a rich repository of high-fidelity, continuous physiological data (ECG/PPG/ABP/ICP) and EHR data, collected from trauma resuscitation units and ICUs. Leveraging advanced machine learning methods, we develop predictive models for critical outcomes in trauma and critical care, such as the need for transfusion, neurological decline, ICU admission, life-saving interventions, and mortality. | We are eager to collaborate with clinical groups possessing similar continuous physiological datasets from pre-hospital, trauma, or ICU settings. Additionally, we welcome partnerships with teams specializing in explainable models, transfer learning, and deep learning to further enhance the impact of our research. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Adam Amos-Binks | Applied Research Associates | aamosbinks@ara.com | Raleigh, NC | ARA health solutions is focused on critical patient care using wearable sensors, physiology simulation, and AI model development. Our AI models employ techniques that help manage risk and adjust to new risks as they emerge. | Clinical expertise in critical care, deployed decision support tools. | TA1: Automated Surrogate Ground Truth Label Extraction |
| Donatello Materassi | University of Minnesota | mater013@umn.edu | Minneapolis, MN | Development of eXplainable AI tools for debugging, anomaly detection, extraction of interpretable terms and interpretable models. Also, system identification and modeling with focus on adaptive systems and robust modeling. | Looking for a research partner with an application that can be powered by AI, but requires explanations for monitoring because of its critical nature. | TA2.1: Continuous Degradation Detection Tools |
| Benjamin Purman | SoarTech | ben.purman@soartech.com | primarily Ann Arbor, MI and Orlando, FL, MI | a. Soar Technology develops AI solutions that reason like and work collaboratively with human. We have over two decades of experience as a trusted provider of AI-enabled solutions across the DoD, supporting domains like: C4ISR, Autonomy, and Training and Simulation. In the health domain, we have developed diagnostic decision aids to support battlefield medics. Additionally, we have extensive expertise in image processing, knowledge elicitation, and machine learning. | SoarTech would be interested in partners with mature health diagnostic tools that could be used to support use case development and dataset development. We would also be interested in talking with potential academic partners. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Reginald Swift | Rubix LS | rswift@rubixls.com | Lawrence, MA | Rubix LS specializes in integrating AI/ML and data analytics into clinical research, focusing on precision diagnostics and personalized treatment solutions. We leverage advanced data platforms and real-world evidence to enhance outcomes in oncology, infectious diseases, and underserved populations. Our expertise spans pre-clinical to Phase II trials, mobile clinical sites, and decentralized patient engagement solutions. | Rubix LS seeks partners with expertise in AI/ML, advanced data analytics, and precision medicine who are experienced in clinical trials, real-world data applications, and patient-centered solutions. We value collaborators with complementary capabilities in software development, bioinformatics, and regulatory pathways, as well as those focused on addressing disparities in healthcare through innovative technologies. | TA2.2: AI-Based Root-Cause-Analysis Tools |
| Weizhi Li | Los Alamos National Lab | weizhi0908@gmail.com | Santa Fe, NM | I am a postdoc at Los Alamos National Lab. I am focused on developing AI technologies to save costs and improve clinical trial outcomes. | I am looking for potential proposers interested in root-cause analysis of clinical AI models, AI model monitoring in clinical trials, and the correction of degraded AI models for clinical purposes. | TA2: Degradation Detection & Self-Correction |
| Genevieve Melton-Meaux | University of Minnesota | Gmelton@umn.edu | Minneapolis, MN | Research focus areas are: (1) Real-world data and real-world evidence generation with data science and AI to further patient care and outcomes; (2) semantic interoperability with FAIR data principles, common data models and terminologies; (3) a leading clinical natural language processing program; and (4) federated learning capabilities. | Complementary data management and AI methodologic capabilities | TA2.2: AI-Based Root-Cause-Analysis Tools |
| Jimeng Sun | University of Illinois Urbana-Champaign | jimeng.sun@gmail.com | Champaign, IL | University of Illinois Urbana-Champaign | Seeking collaborators with expertise in clinical medicine, health informatics and medical imaging. Our team has extensive experience developing clinical predictive models, digital twins for EHR data, and clinical interpretable AI. We offer strong skills in handling large-scale EHR data, model monitoring, and uncertainty quantification. Looking for partners to strengthen clinical validation and multi-institutional data sharing aspects." | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Jinfeng Zhang | Insilicom LLC | jinfeng@insilicom.com | Tallahassee, FL | We specialize in information extraction from natural language and construction of knowledge graphs. We participated in the LitCoin NLP challenge organized by NIH and won first place. We also participated in the BioCreative challenge VIII BioRED track for end-to-end knowledge graph construction and won first place. We have built a large-scale biomedical knowledge graph from all PubMed abstracts, which can serve as a knowledge source to validate AI models. | We are looking for a partner with expertise in AI applications whose degradation detection and/or self-correction can be facilitated by our knowledge graph. We will continuously update our knowledge graph so that it will consist of the latest discoveries and knowledge in biomedical sciences, which can be crucial for continuous detection and self-correction. | TA1: Automated Surrogate Ground Truth Label Extraction |
| Anna Maw | University of Colorado | Anna.maw@cuanschutz.edu | Denver, CO | Our research group has expertise in Implementation Science and Learning Health Systems. We are interested in partnering with AI developers to implement and evaluate processes that optimize AI model performance as well as associated clinical outcomes, health equity, cost, clinician/patient experience across diverse, dynamic, real-world clinical settings over time. We describe an overview of our proposed approach to AI evaluation and maintenance here: (Maw AM et alPMID: 38172408). | We are looking for AI developers interested in implementing and evaluating their AI models in real world clinical settings | TA2.1: Continuous Degradation Detection Tools |
| Eric Eaton | University of Pennsylvania | eeaton@seas.upenn.edu | Philadelphia, PA | I lead the lifelong machine learning research group at the University of Pennsylvania, which focuses on developing systems that are deployed for extended periods, accumulate and refine knowledge over time, and adapt to handle a wide variety of tasks. My group applies these tasks to problems in robotics and precision medicine. Our work combines techniques from continual learning, multi-task and transfer learning, deep learning, reinforcement learning, and interactive AI. | In teaming, we'd be looking for complementary experience, especially on the clinical and informatics side. | TA2.3: AI Model Self-Correction Tools |
| John Kalafut | Asher Orion Group | John.kalafut@asheroriongroup.com | Pittsburgh, PA | We have decades of expertise in medical device development, software engineering, healthcare operations, clinical practice, clinical research, and orchestrating public-private partnerships in medical imaging. Our work in clinical AI monitoring is grounded in principles of quality engineering and includes a reference, process model for clinical AI governance. Specific interest and capabilities are in fairness and robustness assessment, characterization of degradation, and translation science. | We have been assembling a consortium of companies, academics and imaging provider network and looking to further the team structure with expertise in multi-variate dataset characterizations, development of estimation metrics applicable to clinical imaging AI models, and familiarity with dimensionality reduction approaches in image-space. | TA2: Degradation Detection & Self-Correction |
| Pawan Jindal | Darena Solutions | pjindal@darenasolutions.com | St. Louis, MO | Integration of various AI tools into EHRs to align with clinical workflows at the point of care | Organizations that are working on developing AI models that can be integrated into EHRs through APIs | TA3: Quantify Uncertainty & Improve Clinician Performance |
| JEN SIEGELMAN | ASHER INFORMATICS | jen@asheroriongroup.com | Cambridge, MA | Systematic approaches to ensuring imaging AI tool validation and testing in diverse populations | Quantitative partners, clinical partners, government partners | TA2.2: AI-Based Root-Cause-Analysis Tools |
| Vye Greanya | Parallax Advanced Research | viktoria.greanya@parallaxresearch.org | Dayton, OH | Parallax Advanced Research is a non-profit research institute headquartered in Dayton, OH. Current Research Focus Areas: Parallax has scientists with expertise in a general framework for degradation detection and self-correction, including root-cause analysis and metacognitive self-correction. We also conduct research in human factors engineering for understanding uncertainty and improving user performance, including conducting user studies. | SMEs who work on tools for Electronic Health Record data. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Sunil Badve | Emory University School of Medicine | sbadve@emory.edu | Atlanta, GA | I am a surgical pathologist with a broad range of expertise focusing on breast and lung cancers. My Lab is supported by R01 in TNBCs, Komen Scholar award and I collaborate with ECOG-ACRIN and with AI teams. I can assist with clinical and experimental validation of AI -algorithms | Potential partners to develop novel tools for trustworthy AI | TA5: Independent Verification and Validation |
| Danny Chen | University of Notre Dame | dchen@nd.edu | Notre Dame, IN | Dr. Chen has conducted extensive research on image processing and analysis problems in 3D and 4D (3D space plus time) biomedical images. For example, he developed a powerful framework for optimally detecting multiple mutually-constrained boundary surfaces of biomedical objects with complicated topologies (e.g., 3D airway and vascular trees), solving long-standing open problems in computer vision. Further, he has developed many new AI based approaches for various medical image analysis tasks. | Look for medical and clinical experts, who can provide medical data and verify new AI methods in medical settings and applications. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Stephanie Randall | Axle | stephanie.randall@axleinfo.com | North Bethesda, MD | Creating scalable web apps focused on visualization & human-in-the-loop validation on high dimensional datasets (structured/unstructured). Building comprehensive multimodal clinical and pre-clinical biomedical datasets that can be linked using privacy preserving methods. Real world evidence QA/QC, harmonization, causal analysis, and AI/ML powered classification/cohort ID/validation/etc. Leveraging LLMs on unstructured data for element extraction, summarization, knowledge graph creation, etc. | Data providers, linkage experts, secure flexible data analysis platforms, ground truth clinical annotations of unstructured data (images, clinician notes, procedure notes, etc.). | TA4: Core Data Infrastructure |
| Shikha Chaganti | Siemens Healthineers | shikha.chaganti@siemens-healthineers.com | Princeton, NJ | AI in medical imaging | Clinical collaborators | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Steve McNamara | University of Colorado Anschutz | steve.mcnamara@cuanschutz.edu | 1675 Aurora Ct, Aurora, Colorado 80045, CO | Our lab's current research focuses on uncertainty quantification, surrogate ground truth extraction, parsing EHR reports for weak labels using LLMs, and continuous monitoring tools. | We would like to partner with an institution that had a robust core data infrastructure and that has worked on detection of model drift and degradation. | TA2: Degradation Detection & Self-Correction |
| Jodi Wachs | Contexa LLC | contexahealth@gmail.com | Albuquerque, NM | Contexa is developing a domain-wide clinical ontology to address the challenge of structured vs unstructured knowledge in AI tools for CDS. Beginning with feature abstraction and progressing to LLM development, it automates the extraction of surrogate ground truth labels across diverse clinical use-cases. The ontology is continuously updated and revised based on the best evidence, enabling AI models to self-correct and implement updates, ensuring ongoing performance assessment and improvement. | We are seeking to partner with a clinical site and an AI CDS tool provider. Initial discussions have taken place between Contexa LLC and Jaxon.ai, which specializes in trustworthy AI that relies on ground truth data. This collaboration aims to address key challenges, including automating the extraction of surrogate ground truth labels and enabling AI models to self-correct and implement necessary updates, supporting continuous performance improvement. | TA2.3: AI Model Self-Correction Tools |
| Sameer Peesapati | Synthesize | sameer@synthesize.health | Toronto | We work with AI/ML enabled medical technology design and commercialization in the US and Canada. Our work involves understanding the compliance and safety implications of AI based decision support tools built within device or digital products targeting clinicians or use by patients. From our industry experience we bring in a different perspective on what helps with clinical adoption and utilization of AI tools along with paths for reimbursement. | We are interested in collaborating with researchers to build an open source SDK that facilitates self-audit, observability and explainability of AI algorithms - with initial focus on medical imaging and clinical decision support tools | TA2: Degradation Detection & Self-Correction, TA2.1: Continuous Degradation Detection Tools, TA2.2: AI-Based Root-Cause-Analysis Tools, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Ashley Odom | Keywell.ai | ashley@keywell.ai | Austin, TX | HIPAA-compliant enterprise AI infrastructure | We would likely subcontract and support as subject matter experts and partners. | TA4: Core Data Infrastructure, TA2.3: AI Model Self-Correction Tools, TA2.2: AI-Based Root-Cause-Analysis Tools, TA5: Independent Verification and Validation |
| David Kessler | Columbia University Medical Center | dk2592@cumc.columbia.edu | New York City, NY | I have dedicated my career to research in leveraging novel technology to improve training and patient outcomes. I have a broad background in technology-enhanced clinical research, with specific training and expertise in point-of-care ultrasound and healthcare simulation science. As a PI or co-investigator on several large-scale, grant-funded, multi-center studies resulting in numerous publications, I understand the critical importance of constructing a realistic and effective research plan. | As a medical content expert with specific expertise in point-of-care ultrasound, I am looking to join data scientist teams who are seeking subject matter experts for this effort. | TA3: Quantify Uncertainty & Improve Clinician Performance, TA1: Automated Surrogate Ground Truth Label Extraction |
| Julia Komissarchik | Glendor, Inc | julia@glendor.com | Draper, UT | Glendor is on a quest to safeguard patients’ privacy by de-identifying Protected Health Information (PHI) while empowering BAA-free data sharing. Glendor PHI Sanitizer - automatic in situ PHI De-identification software that is easily integrated into the existing data workflow. | We are looking to leverage our expertise in 1. the automatic extraction and integration of data across different clinical use cases to establish a “ground truth” about each patient. 2. aggregation and sharing of data across medical institutions and across performers to advance development | TA4: Core Data Infrastructure, TA1: Automated Surrogate Ground Truth Label Extraction |
| Beatrice Knudsen | University of Utah | Beatrice.Knudsen@path.utah.edu | Salt Lake City, UT | Computational Pathology | Device developers, algorithm developers, telehealth experts | TA1: Automated Surrogate Ground Truth Label Extraction |
| Shahriar Nirjon | University of North Carolina at Chapel Hill | nirjon@cs.unc.edu | Chapel Hill, NC | Robustness, Transparency, and debuggability of AI-enabled sensor technology (e.g., movement, camera, audio, and radio signals) to model human activity and gestures; | (1) Medical domain experts who can bring application-specific requirements; conduct user study; (2) Core machine learning and AI researcher; | TA2.1: Continuous Degradation Detection Tools, TA2.2: AI-Based Root-Cause-Analysis Tools |
| Eric Rosenthal | Massachusetts General Hospital | erosenthal@mgh.harvard.edu | Boston, MA | Our NIH Bridge2AI for Clinical Care Program in the CHoRUS Research Network has developed a 14-center Collaborative Cloud enclave of standardized high-resolution hospital admission data including imaging, waveforms, EHR, and text data, enabling testing algorithms that predict hospital and critical care outcomes longitudinally during a patient’s hospitalization as well as time trends across years, e.g., in response to changes in practice. We maintain teaming, ethicolegal, data, and tooling cores. | AI implementation science skillset, computer science methodological innovation. | TA4: Core Data Infrastructure, TA2.3: AI Model Self-Correction Tools, TA3: Quantify Uncertainty & Improve Clinician Performance, TA1: Automated Surrogate Ground Truth Label Extraction |
| John Sunwoo | Mass General/ Harvard Med School | jsunwoo@mgh.harvard.edu | Boston, MA | Sticker: Physiological waveform analysis for noninvasive monitoring of blood pressure via wearable stickers. We have diffuse optics HW/SW and analysis techniques, including DCS, NIRS, and deep learning. | Potential collaborators would be interested in, with expertise in: Sensor/electronics design/miniaturization, polymer material/science, battery. (Our team poses: clinical team, patient/human study skills, optical techniques, data analysis and deep learning.) | TA1: Automated Surrogate Ground Truth Label Extraction, TA3: Quantify Uncertainty & Improve Clinician Performance, TA4: Core Data Infrastructure, TA2: Degradation Detection & Self-Correction |
| Moh Noori | ScriptChain Health | moh@scriptchain.co | San Francisco, CA, CA | 1. AI prediction models for heart disease readmissions 2. Personalized Medicine including Food as Medicine 3. AI search engine for operational efficiency for medical documents 4. Enhanced digital care coordination 5. Language translation features for improved patient experience | 1. Capital and contracts 2. Healthcare system partners to run studies with 3. Partnerships with industry leaders | TA3: Quantify Uncertainty & Improve Clinician Performance, TA2.3: AI Model Self-Correction Tools |
| Vrad Levering | Triple Ring Technologies | vlevering@tripleringtech.com | Newark, CA | Triple Ring Technologies is a leading partner in developing science-driven products across medtech, life sciences, and sustainability. Our interdisciplinary team, including many PhDs, excels in advancing technologies up the TRL scale and collaborating with academic researchers. We have a strong track record of engaging with ARPA-H, both as subcontractors and primary awardees. We offer services from basic tech development to prototype design and manufacturing. Fully ISO 13485 certified. | We partner with innovators to solve tough problems, launch breakthrough products, and create new businesses. From concept to FDA submission and commercialization, we handle technology development and redesign, as well as complex system integration. We welcome teaming partners with medical or specific domain expertise, and encourage bold ideas beyond current limits. We are experienced at both leading as primary as well as supporting subcontractor roles, and are happy to discuss. | TA2: Degradation Detection & Self-Correction, TA3: Quantify Uncertainty & Improve Clinician Performance, TA4: Core Data Infrastructure |
| Timothy Chou | BevelCloud | tim@bevelcloud.io | Palo Alto, CA | Building accurate AI applications in healthcare requires large amounts of diverse training data. The data is in the buildings – real time data is in the clinical machines and offline data in the PACS and EMRs. Centralized architectures will not work in medicine. The data sizes are much larger, the demands for privacy much higher and the need for real-time results much greater. Instead we have architected a distributed AI infrastructure to move the applications to the data in the clinic/hospital | Groups who have developed deep learning/AI models which were trained on small imaging data sets in all areas of specialization: cardiology, oncology, emergency medicine, orthopedics, radiology, pathology or neurology. Move from the bench to the bedside. | TA4: Core Data Infrastructure, TA2: Degradation Detection & Self-Correction, TA1: Automated Surrogate Ground Truth Label Extraction, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Raghu Machiraju | The Ohio State University | machiraju.1@osu.edu | Columbus, OH | AI-based clinical analyses | Translational applications | TA1: Automated Surrogate Ground Truth Label Extraction, TA2.2: AI-Based Root-Cause-Analysis Tools, TA3: Quantify Uncertainty & Improve Clinician Performance, TA2.1: Continuous Degradation Detection Tools |
| Yiyu Shi | University of Notre Dame | yshi4@nd.edu | Notre Dame, IN | Our team's expertise lies in TA2 and TA3, with focus on on-device AI, domain adaptation, self-supervised learning and model uncertainty and explainability in the healthcare domain. | We are looking for health practitioners to form interdisciplinary teams to address the related clinical challenges | TA2: Degradation Detection & Self-Correction, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Eric Nelson | MORSE | enelson@morse-corp.com | Cambridge, MA | MORSE provides independent testing, evaluation, verification, validation, and assurance for AI decision support tools for US Government customers. We test AI models, including computer vision, language, audio, and multi-modal models. Our research improves IV&V methods, measuring model performance, human-machine interaction, and degradation. We develop automated testing pipelines that adapt to changing conditions, serving as leaderboards for model deployment decisions. | While MORSE has strong experience in AI T&E, we are looking to add team members with specific experience in T&E for X-ray, CT, and/or EHR data. Also connections to medical providers using AI decision-support tools is valuable. | TA5: Independent Verification and Validation, TA2.1: Continuous Degradation Detection Tools |
| Madeline Diep | Fraunhofer USA | mdiep@fraunhofer.org | College Park, MD | Fraunhofer USA is a nonprofit organization and consists of an interdisciplinary team of software and AI engineers. We offer independent assessment service for AI models, and capabilities for targeted augmentation of training and testing dataset. We have developed an AI-diagnostic tool-supported methodology for high-confidence, cost-effective robustness assessments of AI image recognition systems. Our tool can generate test images that incorporate variations expected from biomedical instruments. | We are interested in collaborating with medical partners and developers of AI models, contributing our expertise in assessing and remedying reliability of AI models to clinical decision tool. | TA2: Degradation Detection & Self-Correction, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Pankaj Gore | Insight Health AI | pankaj@insighthealth.ai | Austin, TX | Insight Health has developed an AI governance platform for healthcare applications that involve clinician-patient-AI and patient-AI interactions, and autonomous medical record summarization. We evaluate AI solutions across multiple domains, including but not limited to, data integrity, clinician efficacy and efficiency, patient distress, model resiliency, usability and bias mitigation. Interactions can be risk-scored in real time. The platform is HIPAA compliant and SOC 2 Type 2 accredited. | We are looking for research or healthcare enterprise partners focused on the study of responsible deployment of AI language models in healthcare. Our domain expertise encompasses TA1, TA2, TA3 with a focus on maintaining the safety, trust and ethical norms of healthcare services as AI solutions are implemented. | TA2.1: Continuous Degradation Detection Tools, TA2.2: AI-Based Root-Cause-Analysis Tools, TA1: Automated Surrogate Ground Truth Label Extraction, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Ben Strasser | ASU Research Enterprise | benjamin.strasser@asure.org | Scottsdale, AZ | ASURE (ASU Research Enterprise) specializes in AI development, AI evaluation, metamaterials, and wearable electronics. Foci include translating measures of performance to measures of effectiveness, establishing model-independent user- and theory- driven assessment methodologies, and developing real-time performance evaluation tools for in-situ assessment. These efforts are led by a former DARPA program manager with strong track record in AI evaluation. | Our capabilities include ensuring statistical validity of results and methodologies, validating these results with independent data, and overseeing adherence to engineering and interoperability standards. We seek partners with clinical expertise to ensure clinical relevance of results and interface with clinical sites to enable collection of independent verification and validation data. | TA5: Independent Verification and Validation |
| Matthew Molenda | AM Operating LLC | molenda@anatomymapper.com | Holland, OH | Standards development and maintenance, language processing to ground truth. Anatomy Mapper models make health data AI-ready and research ready by simplifying inputs and automatically coordinating to standards. | Looking for partners to help integrate the Anatomy Mapper technology into various groundbreaking cancer tracking capabilities. | TA4: Core Data Infrastructure |
| Rauf Izmailov | Peraton Labs | rizmailov@peratonlabs.com | Basking Ridge, NJ | From smart cities to smart grids, intelligent battlefields to autonomous systems, our 450+ researchers and engineers are shaping the future—across cybersecurity, electronic warfare, machine learning, analytics, 5G, networking, quantum, and more. | We have solid experience in both fundamental theory of ML/AI and their applications, and we are looking for medical / biological expertise and data access for realizing and calibrating our solutions. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Robert Bock | R-DEX Systems, Inc. | robert.bock@r-dex.com | Woodstock, GA | R-DEX Machine Learning Quality Assurance (MLQA) Tools radically improve the performance and robustness of AI/ML models for a wide range of applications: 1) Lighthouse (Statistical Analysis Package) provides in-depth understanding of dataset quality; 2) Wayfarer (Data Drift and Out of Distribution Detection Package) monitors dataset drift using Stochastic Differential Equations; 3) Bulwark (Adversarial Attack Package) detects, classifies, and corrects for adversarial attacks. | complementary expertise, potentially clinical/healthcare expertise | TA2.1: Continuous Degradation Detection Tools |
| Yanshan Wang | University of Pittsburgh | yanshan.wang@pitt.edu | Pittsburgh, PA | My research lab at the University of Pittsburgh focuses on natural language processing using electronic health records in various clinical applications. | We are looking for partners who have clinical expertise. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Andinet Enquobahrie | Kitware | andinet.enqu@kitware.com | Carrboro, NC | Kitware is a leader in medical imaging AI/ML and computer vision, pioneering advancements in detection, tracking, and scene understanding. Our expertise spans deep learning, multimodal large language models, generative AI, and explainable AI. With deep experience in open-source governance and best practices, Kitware is uniquely positioned to help achieve the PRECISE-AI program's mission of building an open, self-sustaining AI ecosystem. | We are seeking collaboration with clinicians, AI researchers, vendors, and healthcare organizations that can provide deep subject matter expertise, access to clinical data, and insights into real-world healthcare challenges. Ideal partners will bring complementary expertise to enhance AI model development, validation, and deployment within healthcare settings. | TA2: Degradation Detection & Self-Correction |
| Matthew Molineaux | Parallax Advanced Research | matthew.molineaux@parallaxresearch.org | Dayton, OH | Aligned decision-making for medical triage, high-level control of integrated AI systems, metacognitive control | Expertise with medical decision systems and medical decision-making | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Mike Thomas | NobleReach Foundation | mthomas@appianpartnersllc.com | Tysons, VA | Enable transition and translation of innovative research (DARPA, NSF). | Looking for partners in need of transition and translation help for commercialization planning of innovative reasearch. | TA5: Independent Verification and Validation |
| Ju Sun | University of Minnesota | jusun@umn.edu | Minneapolis, MN | automated label extraction; performance monitoring and correction; uncertainty quantification; AI safety | For validation of methods for monitoring and correction; high-quality software implementation | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Antonio Bobbitt | DLH Corp | antonio.bobbitt@dlhcorp.com | 6720B Rockledge Drive, Suite 777 Bethesda, MD 20817, MD | DLH research and artificial intelligence (AI) professionals provide support in epidemiology studies, laboratory processing, and biostatistical analysis. We also specialize in Cloud agnostic Data Services exposed to our customers as a Repository as a Service (RaaS) for search and discovery, retrieval, distribution, domain transfer, and forensics. We apply AI and Machine Learning (ML) to our Extract Transform Load (ETL) processes to ensure accurate data tagging based on our Semantic Ontology. | DLH would like to partner with a firm that leverages our research and data science skills. DLH advances science and knowledge through our extensive research portfolio and domain expertise. We provide large-scale data analytics, clinical trials research services, epidemiology studies, and much more to advance disease prevention methods and health promotion to underserved and at-risk communities. | TA5: Independent Verification and Validation |
| Michael Harman | New England Medical Innovation Center | Michael.w.harman@gmail.com | Providence, RI | Translating science into healthcare solutions. | Core AI model | TA4: Core Data Infrastructure |
| Sanjay Purushotham | University of Maryland Baltimore County | psanjay@umbc.edu | Baltimore, MD | Our expertise lies in AI, Machine Learning, and Federated Learning, particularly in the healthcare domain. Our current research focuses on areas such as Multimodal Generative AI, Medical Image and Video Visual Question Answering (VQA), Explainable AI, Synthetic Data Generation, Privacy-preserving methods, Machine Learning reproducibility and benchmarking. | Clinicians, health science experts, medical organizations | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Federica Zanca | EISMEA | Federica.zanca@ec.europa.eu | Belgium | Medical imaging and AI in healthcare | SMEs or researcher in the field | TA5: Independent Verification and Validation |
| Nate Hughes | Icarus Therapeutics | nate@icarustx.ai | Bay Area and San Diego, CA | Icarus is a CA private HealthTech company with a clinical trials automation platform in enrollment and recruitment that effectively matches patients and physician investigators within a given zip code (including remote enrollment). We solve the supply (e.g. physician investigator or site) and demand (e.g., patient) problem of clinical trials enrollment. | We are looking to build out our team but are very selective with who we bring on board. We need marketing and clinops people. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Christopher Tignanelli | University of Minnesota | ctignane@umn.edu | Minneapolis, MN | I oversee the U.S. federated learning in healthcare collaborative, a collaborative of US Healthcare systems. Additionally we have developed a tool that monitors in near realtime AI model performance following deployment in health systems and can evaluate model drift, performance in specific hospitals and performance across different populations. | Our organization is looking for additional partners to focus on AI model development and validation as well as leveraging the collaboratives solution for monitoring deployed AI solutions. | TA2.1: Continuous Degradation Detection Tools |
| Ouwen Huang | Gradient Health | ouwen@gradienthealth.io | Durham, NC | Gradient Health focuses on creating large scale medical imaging data libraries. Our 100M image library has been used by researchers from Duke University, Harvard, Stanford, and others to solve problems in cancer screening, stroke detection, fracture triaging, LLM development, and many other life saving healthcare technologies. Gradient is used by over 50+ companies which have FDA 510k cleared medical AI algorithms deployed in hospitals and has received support from NIH/NCI. | We are looking for hospital partners which want to join our data network and development partners to help us open-source our de-identification technology. | TA5: Independent Verification and Validation |
| Fartash Vasefi | SafetySpect | fvasefi@safetyspect.com | Grand Forks, ND | Develop AI platform for holistic chronic disease managemnt | How to improve the quality of data in remote areas, indigenous and other minority populations | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Dacre Knight | Mayo Clinic | knight.dacre@mayo.edu | Jacksonville, FL | Clinical outcomes, patient reported outcome measures (PROMs), connective tissue disorders, Long COVID, Neurodivergence, Artificial Intelligence clinical decision support | Data science and engineering, model development | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Kasie Bailey | Truveta | kasieb@truveta.com | Bellevue, WA | Truveta Data studio platform houses over 110 million live, de-identified electronic health records that are research ready, updated daily and available in real time. Our data is inclusive of unstructured data such as clinical notes and images (MRI, Xrays, echos, etc). In addition, Truveta has achieved more than 1.5 million mother-baby linkages supporting unprecedented insights into maternal and pediatric health. Truveta Data also depicts SDOH and can link with other data sets when needed. | Organizations who need live, de-identified RWD. We welcome research organizations and academia as well. | TA1: Automated Surrogate Ground Truth Label Extraction |
| Alexander Nelson | University of Arkansas | ahnelson@uark.edu | Fayetteville, AR | Data infrastructure, super resolution, and multi-modal learning | Model and algorithms researchers or institutions with multi-modal data and learning challenges | TA2.1: Continuous Degradation Detection Tools |
| Salman Asif | University of California Riverside | sasif@ucr.edu | Riverside, CA | My group has been working on algorithms for continual learning, domain adaptation, parameter-efficient model revision/correction methods, targeted unlearning, bias removal, and adversarial learning. | Healthcare experts and partners with complementary expertise and access to clinical datasets. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Tina Bland | Grand Strand Clinical Trial Services, LLC | tinabland@gsclinical.com | Myrtle Beach, SC | Grand Strand Clinical Trial Services is a certified woman-owned small business specializing in supportive services to teams working toward new technology, new drug or new device development. We provide a range of services including but not limited to Clinical Scientist services (i.e. consultative medical writing), proposal development, clinical operations team training and oversight, outsourcing support, data management and QA including in-process and user acceptance testing. | We are seeking to subcontract under a prime provider to offer services within our focus areas. For PRECISE-AI, we can assist in providing the solution summary, pitch deck, and full proposal. We can also support prime teams with independent verification and validation of solutions as well as in providing core data infrastructure to share data across the development teams and advance development for all. | TA4: Core Data Infrastructure |
| Andrew Buckler | BBMSC LLC | andrew.buckler@bbmsc.com | Boston, MA | BBMSC aims to facilitate applications in quantitative systems pharmacology to support various precision medicine workflows for patients and to be used by their care providers. Among these applications include biomarker and surrogate endpoint development and validation, companion diagnostics, therapeutic response simulation, and others. It builds on the convergence of two major technological capabilities: systems biology and the increasing model sizes of AI. | A number of potential partners are considering how to respond and how to organize. I will be attending the meeting with some of them, as it is critical to take a team approach. The meeting will help clarify the best approach to teaming. | TA2.3: AI Model Self-Correction Tools |
| Amanda Ellis | Grand Strand Clinical Trial Services | amandaellis@gsclinical.com | Myrtle Beach, SC | Grand Strand Clinical Trial Services is a certified woman-owned small business specializing in supportive services to teams working toward new technology, new drug or new device development. We provide a range of services including but not limited to Clinical Scientist services (i.e. consultative medical writing), proposal development, clinical operations team training and oversight, outsourcing support, data management and QA including in-process and user acceptance testing. | We are seeking to subcontract under a prime provider to offer services within our focus areas. For PRECISE-AI, we can assist in providing the solution summary, pitch deck, and full proposal. We can also support prime teams with independent verification and validation of solutions as well as in providing core data infrastructure to share data across the development teams and advance development for all. | TA4: Core Data Infrastructure |
| Yi-Chin Tu | Taiwan AI Labs & Foundation | ptt@ailabs.tw | Taipei | Open AI researches partnered with 92% medical centers in Taiwan. | Medical centers and AI researchers. | TA1: Automated Surrogate Ground Truth Label Extraction |
| Théoden Netoff | University of Minnesota | tnetoff@umn.edu | Minneapolis, MN | The Minnesota Personalized Neuromodulation Center (MNPeNCe) is a University of Minnesota - Mayo Clinic partnership dedicated to radically improving neuromodulation outcomes through patient-centered optimization, continuous improvement of optimization techniques, and outreach to the neuromodulation community. We believe that our vision and unique tools will shape all neuromodulation fields in the years to come. | Neurologists and Neurosurgeons that are interested in personalized neuromodulation. Controls engineers in developing novel optimization algorithms for clinical applications. Software engineers to develop clinical grade HIPAA compliant software. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Jeff Anderson | Mayo Clinic | anderson.jeff1@mayo.edu | Rochester, MN | Mayo Clinic Platform is creating a world where care is available to everyon, everywhere. As new technologies create novel opportunities and approaches, Mayo Clinic Platform is using these new technologies to change how care is provided. Mayo Clinic Platform is uniquely positioned to make this ambitious goal a reality because we are founded on Mayo Clinic's dedication to patient-centered care, treatment of rare and complex cases, exceptional outcomes, and world-class research. | Solution developers that add meaningfully to the platform ecosystem we are creating. | TA4: Core Data Infrastructure |
| Christopher Carr | RSNA | ccarr@rsna.org | Oak Brook, IL | Multimodal data collection, de-identification and labeling for AI research focusing on medical imaging. Tooling for imaging data de-identification and processing Extraction of labels from multimodal data sources including through use of foundational models. Incubation of model development through use case-specific challenge competitions. Validation and statistical analysis of model performance, including bias and generalizability assessment and comparison and AI and human performers. | Tooling and expertise for extraction and management of non-imaging clinical variables. Expertise in root cause analysis of model degradation and self-correction methods. Expertise in data gathering for clinician performance. | TA4: Core Data Infrastructure |
| Patrick Connolly | Teledyne Scientific – Durham | patrick.connolly@teledyne.com | Durham, NC | Competency-Aware Machine Learning, Uncertainty Quantification and life-long learning in AI systems, AI-based Image Enhancement | Clinical expertise and workflow integration, Medical Diagnostic and Device Expertise | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Vrad Levering | Triple Ring Technologies | vlevering@tripleringtech.com | Newark, CA | Triple Ring Technologies is a leading partner in developing science-driven products in medtech, that incorporate AI solutions. Our interdisciplinary team, including PhDs and industry experts, excels at collaborating with academic researchers to advance technologies up the TRL scale and to market. Through our deep understanding of the physics of clinical imaging, we have developed in-house tools to simulate accurate and configurable X-ray and CT images that can be used as synthetic data. | We have a strong track record of engaging with ARPA-H, both as subcontractors and primary awardees. We offer product prototyping (ISO 13485 certified) and software development (including AI pipelines and ISO 62304 conforming) services. We are looking for an academic partner that is developing novel techniques in the model explainability (TA3) and/or model degradation detection (TA2) space. | TA2: Degradation Detection & Self-Correction |
| Misa Uno | AI Medical Service Inc. | misa.uno@ai-ms.com | 9th Floor- Chrysler Building, 405 Lexington Ave., New York, NY 10174 USA | Our company specializes in AI software as a medical device for gastroenterological endoscopy, assisting physicians in detecting precancerous and cancerous lesions. We have published the world’s first paper on detecting early gastric cancer using AI and have robust clinical evidence. After obtaining regulatory clearance in Japan, we are now focusing on addressing unmet medical needs and advancing innovation in the US, supported by our new office in New York. | We are seeking clinical research partners to collaborate on advancing AI-based endoscopic technologies, particularly in the detection of early-stage gastric cancer. Ideal partners would be healthcare institutions, research organizations, or clinical networks with expertise in gastroenterology. By working together, we aim to validate our AI system in diverse clinical settings, enhance its performance, and address unmet medical needs in the field of gastrointestinal diagnostics. | TA2.3: AI Model Self-Correction Tools |
| Lauren Sagray | Def-Logix, Inc. | lsagray@def-logix.com | San Antonio, TX | We propose a Computer Network Operations (CNO) co-pilot solution for cybersecurity operations. The copilot is an AI-powered assistant that guides operators through their tasks via step-by-step prompts and suggestions for system testing. It is distinct from a standard LLM in chatbot mode in that it's honed vis a vis prompt engineering and agentic workflow. Our solution ingests text, image and video files, and generates data labels & input analysis to provide one or more courses of action. | Partners with strong healthcare-driven past performance records. | TA2: Degradation Detection & Self-Correction |
| Oliver Aalami | Stanford University School of Medicine | aalami@stanford.edu | Palo Alto, CA | Spezi's current research focuses on developing interoperable digital health applications using modular components that adhere to health data standards like HL7® FHIR®. These modules support data collection from devices, AI model integration, electronic health records, and mobile health systems. | For the PRECISE-AI program, Spezi seeks partners with expertise in advanced AI model development, health data integration, and clinical trial workflows to improve patient outcomes by leveraging personalized, AI-driven approaches. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Guido Mathews | Bayer AG | guido.mathews@bayer.com | Berlin, Germany | Expedite and simplify the AI/ML Algorithm development by addressing challenges in development, validation and deployment, with special focus in the area of medical imaging and data synthesis | Collaborations to improve patient outcomes, improve pharmaceutical industry productivity, making data AI ready and accessible | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Mahmut Kandemir | Penn State | mtk2@psu.edu | State College, PA, PA | I am interested in large-scale data management for ML workloads. I also design hardware accelerators to speed up ML workloads and reduce their energy consumption (and carbon footprint as well) in parallel compute infrastructures. | Expertise in the application areas that are of interest to ARPA-H | TA2.2: AI-Based Root-Cause-Analysis Tools |
| Saied Tadayon | Z Advanced Computing, Inc (ZAC) | saiedtadayon@zadvancedcomputing.com | Potomac, MD | Since 2011, we have been working on a new AI / ML set of algorithms for 3D object/image recognition, using typically only 10 to 50 training samples (instead of 1000s to billions required by Neural Nets used by other companies), with better accuracy & much smaller hardware/ CPU/ GPU, already demonstrated for Aerial Images for US Air Force & Smart Appliances for Bosch. To our knowledge, we are the only company in the world that can train AI using only a few training samples. | World-renowned team of scientists, e.g., Prof Lee, Nobel Prize in Physics & late Prof Zadeh, AI-Hall-of-Fame & Prof Peyman (Inventor of LASIK - awarded a prize by US President). Have over 450 inventions, including 14 issued US patents. Have done 14 invited tech lectures at the US Patent Office on AI/ ML subject (more than any other researcher). Looking for pathologists & other physicians to team up for medical imaging for cancer & many other medical areas. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Heather Whitney | University of Chicago | hwhitney@uchicago.edu | Chicago, IL | I work on developing methods for measuring and evaluating uncertainty in patient-level image quality and AI outputs, including clinician consideration of them. Separately, I also work with MIDRC, the Medical Imaging and Data Resource Center (midrc.org), where part of my efforts are on sequestration of data and private evaluation of algorithms. | For TA3, we are looking for additional sites on which to evaluate uncertainty in algorithms and evaluate clinician performance. For TA5, we are hoping to team with groups interested in working with MIDRC for evaluation of algorithms on sequestered data. | TA5: Independent Verification and Validation |
| Omar Costilla Reyes | Equ Healthcare | omar@equ.care | Boston, MA | Symbolic regression models for artificial intelligence clinical decision tools | Expertise in software as a medical device with a medical and regulatory focus. | TA5: Independent Verification and Validation |
| Ryan McNaughton | PriMetaz | rmcnaughton@primetaz.com | Boston, MA | PriMetaz is an accessory medical device company developing add-on hardware products to improve the signal-to-noise ratio of MRI scans. This allows us to double the SNR of the current clinical standard, which can be translated into improved image resolution and imaging time. We are initiating a first-in-human study to validate our value proposition in a healthy human population. | We are looking for experts in AI/ML with experience automating disease detection from MRI images. We would like to establish the compatibility of our accessory hardware with clinically relevant detection, segmentation, and acceleration software. In this way, we will create a combinatory imaging product to benefit radiologic patient care. | TA5: Independent Verification and Validation |
| Raghu Rayannagari | Singularity IT Solutions LLC | raghu@singularityitsolutions.com | West Chester, PA | Singularity IT Solutions is focused on cutting-edge research in Artificial Intelligence, Machine Learning, and Data Engineering, particularly for healthcare, cybersecurity, and cloud-based applications. We are exploring advanced generative AI models, real-world data integration for personalized medicine, and optimization of algorithms to accelerate drug discovery and cancer research. Additionally, we investigate secure cloud architectures for scalable data processing. | At Singularity IT Solutions, we seek teaming partners with complementary technical expertise in AI, cloud, and data engineering, alongside proven industry experience. We're looking for innovation-driven organizations aligned with our values of integrity, collaboration, and security. A focus on long-term partnerships, regulatory compliance, and cultural fit is essential, as we work to create high-quality solutions for critical sectors like healthcare and government. | TA5: Independent Verification and Validation |
| Chretien Ndindiye | Squid-iQ, Inc. | chretien+arpa@squid-iq.com | Atlanta, GA | SQUID iQ's research is focused on the digital transformation of healthcare technology utilization management (HTUM) to improve efficiency, reduce costs, and enhance patient outcomes. The Squid platform combines accurate, correlated data with workflow automation and device monitoring to eliminate waste and uncertainty and optimize ROA to create "Smart Hospitals" that are more efficient, flexible, and resilient. | We're seeking potential teaming partners with expertise in AI development, data analysis, and clinical applications. These partners should have experience in developing and deploying AI models for various clinical use cases, extracting and integrating data from different sources, monitoring model performance and identifying root causes of degradation, quantifying uncertainty and communicating model output to clinicians, and ensuring the ethical and responsible use of AI in healthcare. | TA5: Independent Verification and Validation |
| Tapan Khan | BioImaginix LLC | khantapan67@gmail.com | Morgantown, WV | Biological markers for neurodegenerative diseases | Diagnostics and therapeutics of neurodegenerative diseases | TA5: Independent Verification and Validation, TA2.2: AI-Based Root-Cause-Analysis Tools, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Philippe Giabbanelli | Old Dominion University | pgiabban@odu.edu | Washington, DC, DC | As director Modeling & Simulation in Health at Old Dominion University, I lead a research portfolio that develops and applies techniques in predictive simulation models, machine learning, and generative AI to problems encountered by clinicians and policymakers. Our portfolio includes a web portal that provide quantitative estimates about suicide prevention policies in the US (deployed within the CDC) and several state-of-the-art models in fields such as obesity prevention. | I am interested in practical challenges where the technical expertise of my group in computer science and our years of experience in health informatics can lead to fruitful collaborations with the needs of health practitioners. We are ready to collaborate with planners and public health policymakers (particularly in areas such as mental health and chronic diseases) and we are expanding our portfolio in decision-support tools for clinicians. | TA2.2: AI-Based Root-Cause-Analysis Tools, TA2.3: AI Model Self-Correction Tools, TA5: Independent Verification and Validation |
| Sudeepta Mondal | RTX Technology Research Center | sudeepta.mondal2@rtx.com | East Hartford, CT | RTRC is the corporate research division of RTX. RTRC is actively involved in several internally and externally (e.g. DARPA, ARPA-E, IARPA, AFRL) funded research activities related to safety-critical applications with a strong focus on data and model drift detection, safe and robust AI deployment. | Of relevance to the ARPA-H PRECISE-AI program, we have developed tools for data and model drift detection, mitigation, continuous monitoring/learning starting from fundamental research, up to high TRL-level products in a variety of safety-critical domains. These methods have been applied to multi-modal data, such as images, text, audio and time series, with large-scale computer vision and Generative AI/LLM model architectures. We have multiple patents covering our contributions in this area. | TA2: Degradation Detection & Self-Correction, TA2.1: Continuous Degradation Detection Tools, TA2.2: AI-Based Root-Cause-Analysis Tools, TA2.3: AI Model Self-Correction Tools |
| Sardar Ansari | University of Michigan | sardara@umich.edu | Ann Arbor, MI | The Data Science Team at the Weil Institute at the University of Michigan has been developing methods for post-market surveillance of clinical AI/ML tools. In particular, the team has been focused on detection of dataset shift in the electronic health records caused by numerous sources of variation in clinical data. In particular, the team is leveraging causal inference and counterfactual analysis to address the fundamental issue of drift detection in clinical prediction models. | We hope to partner with other teams who have access to data from previous, current or future clinical trials to enable the validation of degradation detection tools. | TA2.1: Continuous Degradation Detection Tools |
| Karen Duvall | Tata America International corp. | Karen.duvall@tcs.com | Milborn, OH | AI and it's many potential uses | Small businesses of various socio-economic types with past performance in one or more of the technical areas | TA1: Automated Surrogate Ground Truth Label Extraction |
| Prativa Hartnett | Southwest Research Institute | prativa.hartnett@swri.org | N/A, TX | SwRI is a non-profit research organization that focuses on developing cutting-edge AI/ML technologies in the healthcare field. Our team's expertise includes Natural Language Processing techniques to extract insights form large amounts of text data, such as EHRs. Our computer vision team research image analysis algorithms to detect patterns and anomalies in medical images such as X-rays, CT scans and MRIs. We also focus on XAI to increase transparency/explainability in healthcare modeling. | Overall, the SwRI team has expertise in data science, machine learning and artificial intelligence. For this program, SwRI is interested in collaborating with organization who have FDA approved clinical decision support tools to partner with in mitigating degradation, continuous assessment and root cause analysis. We are interested in collaborating with subject matter experts who would complement our capabilities. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Mahnaz Maddah | Broad Institute of MIT and Harvard | maddah@broadinstitute.org | Cambridge, MA | ML4H ( https://www.broadinstitute.org/ml4h ) at the Broad Institute is a team of ML scientists who, in collaboration with faculty and clinical researchers from Massachusetts General Hospital, Brigham, and Women’s Hospital, develop advanced clinical AI methods and applications. | We are interested in collaborations in which we could leverage our expertise in clinical AI method development and testing the methods in the large longitudinal clinical cohorts of MGH and BWH. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Ishanu Chattopadhyay | University of Kentucky | ishanu_ch@uky.edu | Lexington, KY | My research group focuses on the development, deployment and validation of cutting edge AI for screening, diagnosis and prognostic prediction of complex diseases, based on EHR data. Full publication list is available at https://zed.uchicago.edu/publications_by_type.html Representative publications for TA1 and TA2 are: https://www.nature.com/articles/s41591-022-02010-y and https://royalsocietypublishing.org/doi/10.1098/rsif.2014.0826 | Implementation collaboration. Other data types beyond EHR | TA2: Degradation Detection & Self-Correction |
| Neena Imam | SMU | nimam@smu.edu | Dallas, TX | High performance computing, privacy preserving machine learning | health data analytics | TA2.3: AI Model Self-Correction Tools |
| Annie Soh | AIMIS | anso88@protonmail.com | Tampa, FL | AI-enabled Medical Devices and AI Software as Medical Devices (SaMD) Independent Verification & Validation, Technical Review / Testing | As an AI Technology expert with specific expertise in AI Testing & Evaluation (T&E), Independent Verification and Validation (IV&V) of AI Models / Systems operational performance, I am looking to join data scientist or AI engineer teams seeking subject matter experts in AI Testing, Independent Verification and Validation efforts. | TA5: Independent Verification and Validation |
| Farinaz Koushanfar | University of California San Diego | fkoushanfar@ucsd.edu | La Jolla, CA | Our research is heavily focused on safety, reliability, privacy, and security of the emerging AI models and systems. | We are looking to partner with various entities with access to various data sets. | TA1: Automated Surrogate Ground Truth Label Extraction |
| Axel Wismueller | University of Rochester Medical Center | axel.wismueller@gmail.com | Rochester, NY | We have invented, tested, evaluated, and published novel methods for quantitatively evaluating performance and usefulness of AI solutions in the clinical practice of radiology, such as in prospective randomized virtual clinical trials (AI-PROBE, method has received ACR Innovation Award), re-defining radiology quality assurance programs using AI-based degradation detection (AQUARIUS), and multi-institutional data harvesting (MIDH) for large-scale AI-based nation-wide healthcare data analysis. | We are looking for experienced academic and commercial partners for public-private partnership collaborations, with the goal to develop innovative methods for quantitatively evaluating the real-world usefulness of AI solutions in clinical practice. As the PI is both a clinically practicing radiologist and a machine learning scientist, our goal is to develop, deploy, and evaluate large-scale AI-based image interpretation, clinical understanding, and efficient health-care management methods. | TA4: Core Data Infrastructure |
| David Schroh | Uncharted Software | dschroh@uncharted.software | Toronto, Canada | Uncharted is actively engaged in research into uncertainty quantification methods for AI systems, improving user outcomes by developing innovative ways to represent model uncertainty, and novel measures that enable users to effectively leverage uncertainty models for better decision-making. Uncharted has a proven history in R&D for human-in-the-loop processes and analytics, building and enhancing user trust in AI systems, and modeling, measuring, and improving user performance with AI. | Uncharted brings a wealth of experience as a standout performer in DARPA programs, working across a broad variety of domains from health to human security. We have a strong track record working with many ARPA-H leaders. Uncharted is looking for collaborative partners who are passionate about highly creative research that can have groundbreaking, real-world impact, and that can extend our reach to more clinicians and clinical settings, and in performing field studies. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Andrew Omidvar | Philips | andrew.omidvar@philips.com | Washington, DC | AI for healthcare, medical devices and clinical solutions for prevention, diagnostic and treatment of diseases in cardiology, oncology and neurology. With scale up capability and impact in increasing access to care with sustainable business model. | Companies with capabilities for use of AI in medical devices such as wearables. | TA4: Core Data Infrastructure |
| Aakriti Pandita | University of Colorado | aakriti.pandita@cuanschutz.edu | Denver, CO | Safe and effective LLM implementation in healthcare | NLP scientist with deep technical skills looking to collaborate on bold ideas | TA5: Independent Verification and Validation |
| rodger kessler | university of Colorado Anschutz School of Medicine | rodger.kessler@cuanschutz.edu | Denver, CO | The IRIS focus is infrastructure support for multi on the ground interoperable support for an algorithm that predicts patient risk of poor medical and functional outcomes and implement in multiple independent family medicine practices using 4 different non EPIC EHRs, that have limited IT support. The risk prediction will be accompanied by work flows at POC to respond to modifiable risk elements. Data sources include EHR, utilization, and place data, and patient reported Quality of `Life. | MLOps/DevOps; AWS expertise, algoritm development, model tuning of hyperparameters, CI/CD for ML modeling, code repository and version control interface with Git. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Oleg Pianykh | MGB, Harvard Medical School | opianykh@mgh.harvard.edu | Boston USA, MA | Stability of AI models in healthcare. This includes identifying the hidden causes of instability, developing stability metrics, developing applied and mathematical approaches to stable AI | Interested in collaborating with applied mathematicians, AI developers, healthcare practitioners using AI. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Martin Willemink | Segmed | martin@segmed.ai | Palo Alto, CA | Segmed developed the technology to connect to IT systems for medical imaging (PACS), de-identify text and pixel information, standardize imaging data, and we built a platform to search and filter through imaging data. | We can provide data and data curation tools for medical imaging, but are not focused on the other areas | TA4: Core Data Infrastructure |
| John Powderly MD | Carolina BioOncology Institute, PLLC & BioCytics Inc | jpowderly@carolinabiooncology.org | Huntersville, NC, USA, NC | Carolina BioOncology Institute (CBOI) PLLC is a phase 1 cancer research clinic, located adjacent to BioCytics Inc Human Applications Lab's GMP cell processing facility. Both companies were founded in 2005 by oncologist Dr. John Powderly with vision to enable autologous immune cell therapies manufactured at the point of care for rapid iteration and clinical delivery. BioCytics Digitally Integrated Clinical Enterprise (BioDICE) enables real-time EMR-EDC-HALIMS-CTMS single source data for AI. | CBOI-BioCytics hosts 19 years digital history of metastatic cancer patients' treatment history and symptom management linked with large biobank of autologous tumors and matched pair immune cells for clinical-bioinformatics foundational data for LLM and AI projects. Looking for expert partners with experience building AI from large data sets focused on cancer to integrate AI in future BioDICE software development to help patients & oncologists develop, manufacture & deliver cell therapies. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Daniel Bryce | SIFT, LLC. | dbryce@sift.net | Minneapolis, MN | Smart Information Flow Technologies (SIFT) is a research and development consulting company with strengths in Natural Language Processing (NLP), Automated Planning, Cybersecurity, Supervisory Control, Healthcare, and a range of Human-Automation Interaction technologies. SIFT Researchers specialize in Artificial Intelligence, Software Engineering, Linguistics, Control Theory, Neuroscience, Human Performance, and Politeness and Etiquette models. | SIFT is seeking partners in TA1 and TA3, especially in terms of clinical expertise. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Samuel Scarpino | Northeastern University | s.scarpino@northeastern.edu | Boston, MA | Development of explainable, responsible, human-in-the-loop AI solutions | Data infrastructure partners, root-cause-analysis partners, and clinical partners | TA5: Independent Verification and Validation |
| Eric Rosenthal | Massachusetts General Hospital | erosenthal@mgh.harvard.edu | Boston, MA | Working with sites through the Bridge2AI for Clinical Care CHoRUS Program, we develop AI-ready datasets for clinical care AI by harmonizing imaging, EHR, and waveform data from 14 hospitals in a central collaborative cloud environment with tools for machine learning, interpretability, and performance assessment. | Expertise in federated learning | TA1: Automated Surrogate Ground Truth Label Extraction |
| Stephen Applebaum | EM Neuroimaging | sia@emneuroimaging.com | Phoenix, AZ | EM Neuroimaging is currently a small organization that performs consulting services for MEG and EEG companies and users. We are partnering with others to explore how generative AI can be used both to develop MEG/EEG analysis techniques and improve the underlying software itself. | Anyone interested in functional neuroimaging, particularly in clinical workflows. | TA4: Core Data Infrastructure |
| Charlotte Kalafut | Asher Informatics PBC | charlotte@asherinformatics.com | Pittsburgh, PA | Asher Informatics PBC provides Medical Imaging AI Quality Assurance solutions, empowering safer and more effective AI Adoption through a suite of pre-deployment fairness audits and post-deployment monitoring solutions. Our R&D interests and capabilities in subgroup and covariate analyses of medical imaging to identify and mitigate hidden stratification, applications of algorithmic fairness to the regulated medical domain, image quality assessment, and multi-variate data drift analysis. | Asher Informatics PBC brings background IP consisting of hybrid-cloud methods applicable for assessing data and AI performance in healthcare IT, production settings. We are looking to bring our domain knowledge in medical imaging (physics, informatics, and medical) to teams seeking SMEs and where our capabilities could enhance theirs. | TA2: Degradation Detection & Self-Correction |
| Paula Huston | MITRE | phuston@mitre.org | Bedford, MA | MITRE’s mission-driven teams are dedicated to solving problems for a safer world. Through our public-private partnerships and federally funded R&D centers, we work across government and in partnership with industry to tackle challenges to the safety, stability, and well-being of our nation. MITRE offers a range of capabilities and technical expertise for health AI. | TBD | TA5: Independent Verification and Validation |
| Akshay Sood | GrammaTech | asood@grammatech.com | Ithaca, NY | We have a strong background in AI applications to clinical informatics settings, including expertise in several relevant areas: - Extracting multimodal features from EHRs - Unsupervised learning over EHRs - EHR-based risk prediction models for clinical decision support - Model explanation in multiple domains, including clinical risk prediction and malware detection - Synthetic data generation - AI/LLM applications to software analysis and security - Continuous software monitoring tools | For the PRECISE-AI program, we are looking for clinical experts with real-world experience in deploying decision support tools in clinical settings, as well as clinical researchers with experience in the use of predictive analytics and machine learning applied to clinical data, such as for enhancing patient care and operational efficiency, in a range of patient and healthcare environments. We would also welcome partnering with other AI researchers with expertise complementary to our own. | TA1: Automated Surrogate Ground Truth Label Extraction |
| Siun-Chuon Mau | CACI, Inc. - FEDERAL | siun-chuon.mau@caci.com | Florham Park, NJ | AI/ML technologies, including ones we believe are applicable to TA1 and 3. | Clinical partners in possession of multi-modal EHR with ground-truth diagnoses and/or with capacity for field studies assessing clinical efficiency. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Sharon Ricks | Ricks Consultants | sharon.ricks@outlook.com | Austin, TX | Our focus is on building ethical AI by incorporating stakeholder engagement. We can assist ARPA-H applicants in developing and testing an ethical, community-driven AI framework specifically designed to address health disparities. Our aim is to ensure that organizations adhere to ethics principles to ensure that their applications are safe, secure, and reliable. | We are looking for partners who need help developing and utilizing an ethical AI framework that ensures that AI algorithms don't perpetuate existing biases and discrimination (by being trained using biased data). We also seek partners who are serious about accountability for the decisions made by the system, explainability, transparency, farness. | TA4: Core Data Infrastructure |
| Jackie Cha | Clemson University | jackie@clemson.edu | Clemson, SC | human factors; ergonomics; surgery; robotics; exoskeletons; usability; physiological and behavior sensing | Healthcare collaborations | TA3: Quantify Uncertainty & Improve Clinician Performance, TA2.2: AI-Based Root-Cause-Analysis Tools, TA2: Degradation Detection & Self-Correction |
| Sanjay Purushotham | University of Maryland Baltimore County | psanjay@umbc.edu | Baltimore, MD | Our expertise lies in AI, Machine Learning, and Federated Learning, particularly in the healthcare domain. Our current research focuses on areas such as Multimodal Generative AI, Medical Image and Video Visual Question Answering (VQA), Explainable AI, Synthetic Data Generation, Privacy-preserving methods, and Machine Learning reproducibility and benchmarking. | Clinical experts, healthcare organizations, health science domain experts | TA2: Degradation Detection & Self-Correction, TA3: Quantify Uncertainty & Improve Clinician Performance, TA1: Automated Surrogate Ground Truth Label Extraction |
| Keith Dufendach | University of Pittsburgh Medical Center | Dufendachka@upmc.edu | Pittsburgh, PA | Translating artificial intelligence and machine learning technologies to improve patient care in cardiovascular and critical care. | Machine learning experts interested in translating AI/ML technologies into healthcare, particularly in the areas of cardiovascular and critical care medicine. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Sharon Li | UW-Madison | sharonli@cs.wisc.edu | Madison, WI, WI | My research lab has been focusing on the algorithmic and theoretical foundations of reliable machine learning, addressing challenges in both model development and deployment in the open world with evolving and unpredictable data. We develop principled uncertainty estimation algorithms for deep neural networks, and have explored clinical applications in the past. | Experience in bridging deep learning algorithms and clinical settings. | TA3: Quantify Uncertainty & Improve Clinician Performance |
| Kyle Bergquist | Reputable Health | kyle@reputable.health | San Diego, CA | Reputable Health focuses on leveraging real-world data and AI-driven analytics to support product validation, compliance, and performance optimization for health solutions. Our platform integrates wearables and advanced data analysis tools to enhance study design, identify trends, and validate health claims, ensuring regulatory compliance and actionable insights for product development and market readiness. | Ideal teaming partners for Reputable Health include organizations specializing in AI-driven health analytics, wearable technology developers, and regulatory compliance experts. Collaborations with academic institutions or research hospitals for clinical trials and data validation are also key. Additionally, partnerships with companies focused on healthcare data security and cloud infrastructure would ensure robust data handling, supporting large-scale studies and regulatory submissions. | TA2.3: AI Model Self-Correction Tools |
| Ed Middleton | Coalition for Health AI | ed@chai.org | Boston, MA | Certification of Assurance Labs: Supporting the development of a nationwide network of quality assurance labs to evaluate and validate AI models. Continuous evaluation: Providing tools and frameworks for ongoing testing and monitoring of AI models to ensure they remain safe, effective, and unbiased over time. Best Practice Guidance: With the input of experts, developing best practice guidance, use case by use case. A publicly-available national registry. | CHAI is assembling an experienced and diverse partnership of hospital providers, AI-Decision Support Tool developers, and AI Assurance Labs. CHAI will bring together a range of senior clinical and technical experts that can agree approaches and standards for Technical Areas 1-3. CHAI is keen to discuss the proposal with any organization falling into the above three categories, who are interested in collaborating. | TA2: Degradation Detection & Self-Correction |
| Tien Hsin | CareCost AI | tyhtienyu@gmail.com | Boston, MA | CareCost AI an AI agent that helps you shop for healthcare and it’s your one stop shop for everything you need to know about any health services | Veteran in healthcare system, AI experts, and beta users | TA2.3: AI Model Self-Correction Tools |
| Joshua Eggleston | huru | josh@hurulabs.com | London, United Kingdom | Developing the worlds first self-applied clinical grade extended duration neuromonitoring devices. Our research focuses on miniturization of technology, improving signal quality and extending monitoring durations. We are also working on user-friendly designs for patient comfort and ease of use. Our efforts include clinical validation studies, reglatory compliance and developing machine learning algorithms for data analysis to improve the diagnosis and management of neurodegenerative conditions | huru seeks partners with a focus on AI capabilities. An ideal partner would be a company or organisation able to take a rich continuous new dataset and build the models on top of it. | TA5: Independent Verification and Validation, TA3: Quantify Uncertainty & Improve Clinician Performance |
| Sudeep Hegde | Clemson University | sudeeph@clemson.edu | Clemson, SC | Complex adaptive systems; human-AI teaming; safety report analysis; proactive learning of adaptive strategies; decision support; natural language processing; analyzing data at-scale to identify patterns of coordination and adaptive decision making with trade-offs (e.g. efficiency vs. thoroughness) in complex systems | Machine learning and AI expertise; access to healthcare data (EHR data, hospital incident reports, progress notes etc.); expertise in clinical domains (e.g. radiology, emergency medicine, critical care). | TA3: Quantify Uncertainty & Improve Clinician Performance |