ARPA-H Project Awardees
The ARPA-H Mission Office Innovative Solutions Openings (ISOs) and Open Broad Agency Announcement (BAA) provide funding for research that aims to improve health outcomes across a wide range of patient populations, communities, diseases, and conditions. These projects focus on transformative ideas for health research breakthroughs or technological advancements.
Awards made from the ISOs and Open BAA are generally in the form of contractual agreements. Exact award amounts are dependent upon meeting milestones typical of the ARPA-H process. As of March 2024, ARPA-H is no longer accepting submissions for the Open BAA solicitation, but ARPA-H will continue to review and consider solution summaries and proposals submitted under the Open BAA before it closed. Currently, ARPA-H will primarily use Program-Specific and Mission Office ISOs to advertise and accept submissions for its programs and projects.
ARPA-H is pleased to announce the following awardees:
AVOID-OME: Structurally enabling the “avoid-ome” to accelerate drug discovery
Small molecule therapies represent a significant portion of FDA-approved drugs, yet their development is often hindered by the challenge of balancing on-target activity with desirable pharmacokinetic (PK) properties. Pharmacokinetic properties, which encompass absorption, distribution, metabolism, and excretion (ADME), determine a drug's fate within the body and are crucial for its efficacy and safety. Currently, the optimization of these properties is often addressed late in the drug discovery process, leading to costly late-stage failures. This project seeks to overcome this limitation by proactively characterizing the chemical space accessible to ADMET-associated proteins ("anti-targets"). By applying recent advances in experimental and computational techniques, a comprehensive open library of experimental and structural datasets will be generated. This precompetitive resource will provide valuable insights into the binding properties of "anti-targets" and empower researchers to develop predictive AI models for pharmacokinetic optimization. This shift towards a more proactive and data-driven approach promises to significantly expedite drug discovery by streamlining lead optimization, mitigating late-stage attrition, and ultimately accelerating the delivery of new therapies to patients.
TARGET: Transforming Antibiotic R&D with Generative AI to stop Emerging Threats
The Transforming Antibiotic R&D with Generative AI to stop Emerging Threats (TARGET) project will use AI to speed the discovery and development of new classes of antibiotics. Bacterial infections are a leading cause of death worldwide, and there is an urgent need to develop new antibiotics as the prevalence of antibiotic-resistant bacteria grows. Conventional efforts to identify and develop new antibiotics require extensive manual screening and testing of molecular compounds, with the majority failing to successfully show antibiotic activity. This laborious process impedes our ability to discover new antibiotics at the speed needed to address the urgent threat of antimicrobial resistance. TARGET aims to close this gap by using deep learning to identify biomolecules with antibiotic and pharmaceutical potential and using generative AI to broaden the pool of candidate molecules. Together, these approaches will speed the development of new classes of previously unknown antibiotics.
Cancer Identification and Precision Oncology Center (CIPOC)
This project aims to create Oncology Learning Health Systems among health systems of all sizes and at any location. The overall scope of the University of North Carolina (UNC)-based center is to bring together structured and unstructured electronic health record (EHR)-derived data, insurance claims, and geographic data to rapidly identify cases and improve care quality via access to cutting edge clinical advances for cancer. The Cancer Identification and Precision Oncology Center (CIPOC) will facilitate an Oncology Health Learning System, through rapid, real-time identification and accurate characterization of newly diagnosed cancer cases for surveillance, patient recruitment, optimization of patient care, and largescale population research, providing equitable access to precision medicine.
SQUEEZES: Sharing Data for Rapid Collaboration on Encrypted Data
The SQUEEZES framework accelerates the sharing and collaboration of highly sensitive patient data, enabled by emerging technologies for Privacy Enhanced Computations (PECs) and Privacy Preserving Record Linkage (PPRL). This framework utilizes PECs to expedite collaborative data access and enables researchers to build higher quality models from richer aggregated data linked and joined with PPRL from multiple sources for centralized data analysis and management. SQUEEZES is particularly vital when it is difficult or impossible for data owners to share their data due to privacy concerns, allowing them to retain control over how their data is used. It is a standards-compliant framework designed to: (1) reduce the timeline of sharing data from months to hours while lowering financial and administrative barriers for data sharing, (2) automate the sharing and linking of privacy-protected data used in analytics workflows, reducing the workload of data alignment and schema matching, (3) support existing cancer research data workflows, (4) allow data owners to control the use of their data even after it is shared, protecting patients and data owners, including historically underserved communities, (5) create an incentivization framework for data owners to maintain control over the results of computations on their data, and (6) be scalable, extensible, and easily deployable on legacy and emerging standards-compliant systems.
CAIDF: Creating AI-enabled All-Health Team Data Fabric
The objective of the project is to create scalable solutions to unlock previously underutilized nursing, physical therapy, occupational therapy and speech and language pathology (RN/PT/OT/SLP) data with routinely collected medical data and integrate it with AI technology. This project focus on RN/PT/OT/SLP data, which is updated more frequently and regularly than data from physician providers in both inpatient and outpatient settings. Specifically, this project will target two common, costly patient populations from across the lifespan whose care needs require multiple disciplines: 1) adult patient falls with trauma/injury requiring hospital admission, and 2) the transition from the neonatal intensive care unit (NICU) to home with preterm or medically complex neonates. The team will use human-curated AI to enable access to the data and innovate on the summarization of the complex episodes of care. The project performance sites include one urban hospital serving a diverse metropolitan population and two hospital systems in smaller cities, each with a large rural catchment area, which will provide racial and ethnic diversity, as well as rural representation in our enclave data. The multidisciplinary LLMs will be released to the world at the end of this project to drive future innovation.
VECTORS
The VECTORS project aims to create a rapid, safe, scalable, and low-cost plant-based manufacturing platform to drastically reduce production time and cost of viral vector-based gene therapies. Cirsium Biosciences leverages plant-based manufacturing (upstream and downstream) to reduce both production costs and time. Cirsium’s goal is to establish and validate a highly scalable, reproducible, cost effective, rapid plant-based viral vector production platform. Plants have been used to produce biologics for decades and are a proven production host for biologics that have successfully entered clinical trials and been approved by regulators. Cirsium’s process already achieves up to 80% reduction in manufacturing lead times for AAV gene therapies with room for substantial improvements and optimizations. If successful, the plant-based production of viral vectors used for gene therapies will help address the increasing demand for these vectors and the looming shortages for some vector types.
No Kidney Left Behind
34 Lives seeks to rehabilitate approximately 50% of the otherwise discarded donor kidneys back into the donor pool by using Normothermic Machine Perfusion (NMP) technology to enable real-time recovery of organ viability and function. The NMP and other novel work around rehabilitating kidneys that are discarded, diseased, or ischemically injured will be done at a core preservation hub operating as CLIA- certified clinical laboratory in association with designated Organ Procurement Organizations (OPOs).
Improving the patient experience: LLMs for results interpretation and discharge paperwork
The goal of the project led by Stanford University is to develop a fully automated and scalable approach for optimizing large language model medical chatbots as well as clinician-grade evaluation of the results, for patient-facing use cases. Immediate use cases include the evaluation of a chatbot to help patients understand imaging results and a chatbot to aid patients in understanding instructions after hospital discharge.
ALISS: Autonomy at a Less Invasive Scale in Surgery
This project aims to create a full set of modular tools for autonomous robotic surgery, as well as the perceptual capabilities and logical framework to integrate them into full procedures. The team has divided surgical procedures into seven core tasks: retraction, resection, hemostasis, proprioception, debridement, palpation, and suturing. The team then creates a logical framework that defines procedures in terms of these tasks and allows the robot to progress from one task to another. As the robot carries out a procedure, it not only assesses completeness of the subtask, but also its own uncertainty, allowing it to request human surgeon intervention when uncertainty is high. Development is centered around the Virtuoso, a surgical robot with a novel concentric tube form factor and narrow dimensions that allows it to bend and reach difficult anatomical locations that neither current endoscopes nor human surgeons can easily access. Although the tool attachment and the perceptual systems are specific to the Virtuoso, the task algorithms and logical framework will be robot agnostic. The technology developed in ALISS will endow the Virtuoso robot with autonomous surgical capabilities and advance the state of the field by developing generally applicable logical frameworks and algorithms.
MEDIC: Monitoring, Evaluation and Diagnosing Intelligent Chatbots
The goal of the project led by Raytheon BBN Technologies is to develop an overall process for the evaluation of patient-facing medical-domain chatbots, covering a wide range of metrics and grounded in medical practice by the elicitation of desires and concerns from medical staff, caregivers, and patients. This project will use state-of-the-art large language model capabilities piloted across multiple US Government-funded projects to augment and enrich an initial starter-set of data to cover the problem space and use a collection of Machine Evaluators to produce the metrics, with a minimum of human supervision. The final deliverable will consist of a pipeline of tools for prompt enrichment and a similar system for chatbot evaluation. Immediate use cases include prenatal care in the first and second trimester and mental health in young adults.