PMI’s mediKanren tool rapidly identifies FDA-approved drugs as potential treatments for rare and complex diseases by analyzing millions of scientific papers and pharmaceutical databases.
This innovative tool has identified therapeutics for two clinical trials, including a breakthrough trial for ADNP-driven autism, where low-dose ketamine became the first AI-predicted therapy to be clinically validated. mediKanren uncovers new ways to target genes linked to diseases, and vetted therapeutic options are shared with physicians for clinical consideration. UAB is also developing the user interface for the NIH’s NCATS Biomedical Data Translator, making this tool publicly available. Researchers and patient foundations can collaborate with PMI to create custom drug repurposing lists, offering new hope for the 95% of rare diseases without FDA-approved treatments.
Generative AI for Disease Diagnosis and Treatment
PMI has recently been expanding the capabilities of its flagship mediKanren AI tool with generative AI capabilities: mediKanren-GPT is a new AI assistant leveraging the billions of curated biomedical facts within mediKanren that can assist clinicians with disease diagnosis and treatment. Because generative AI excels at extracting facts from scientific papers, mediKanren's mediKanren-GPT will also significantly expand the curated facts within mediKanren's knowledge base. These combined capabilities already enable mediKanren-GPT to adopt a specific medical persona – such as infectious disease, addiction medicine, genetics, or cardiology – to provide detailed information about disease diagnosis and treatment.
ARPA-H initiatives
PMI is leading several groundbreaking ARPA-H initiatives aimed at revolutionizing biomedical science and patient care through AI and data integration. In collaboration with Harvard, UNC, and Maastricht University, PMI is developing CHARM, a toolchain designed to harmonize and analyze biomedical data, unlocking its potential to accelerate discoveries and personalized treatments. Additionally, PMI is partnering with the Every Cure Foundation on the MATRIX project to use AI to repurpose existing drugs for rare diseases lacking effective therapies. Through the HealthyDocs project, in collaboration with Galois, Inc. and the University of Colorado, PMI is advancing secure healthcare data management, facilitating research and care for rare genetic disorders while safeguarding patient privacy.
Associated Studies
Partners
NCATS Biomedical Translator Program
Harvard University
University of North Carolina
Maastricht University
Every Cure
Every Cure aims to alleviate suffering for 300 million people worldwide by maximizing the potential of existing drugs to treat as many diseases as possible, offering a rapid path to saving lives.
Galois, Inc.
University of Colorado