Ó³»­´«Ã½

Cardiovascular Disease Initiative (CVDi) at the Ó³»­´«Ã½ 

The ML4H initiative partners closely with the Ó³»­´«Ã½'s Cardiovascular Disease Initiative (CVDi), led by Patrick Ellinor, to advance cardiovascular research by leveraging machine learning. This long-standing collaboration brings together ML4H’s technical expertise in computational modeling and large-scale biomedical data analysis with CVDi’s deep domain knowledge in cardiovascular biology and genetics. Joint efforts span developing predictive models for heart disease risk, interpreting complex genomic data, and translating insights into clinical utility. 

Together, ML4H and CVDi aim to accelerate scientific discovery and improve cardiovascular health outcomes through innovative, data-driven approaches. This collaboration has led to 40+ publications in top-tier journals and the initiation of prospective studies at Mass General Brigham (MGB) using the developed models. 

MGB Division of Clinical Research

ML4H is engaged in a multi-year collaboration with the to advance the development of a platform that enables clinical researchers to easily query rich patient-level data and apply machine learning analyses without technical barriers. To support this vision, ML4H is building a scalable infrastructure on MGB’s Azure cloud, beginning with the 400,000-patient Primary Care Linkage Cohort. Over successive phases, the team has developed key components including a deep learning pipeline for clinical event detection, generative AI tools for summarizing clinical notes, and the ATLAS interface for subcohort creation and exploratory analytics. 

By combining cloud-native infrastructure with cutting-edge ML methods, this partnership is transforming how clinical researchers across MGB engage with data, formulate hypotheses, and accelerate discoveries in real-world settings. 

The NIH All of Us Research Program

As part of a federally funded initiative through the , ML4H is leading a transformative effort to harness longitudinal wearable sensor data and multimodal AI to better understand real-world patterns of physical activity, sedentary behavior, sleep data, and their impact on health across the U.S. population. Led by Mahnaz Maddah as principal investigator, this work integrates large-scale wearable data, electronic health records (EHRs), and other modalities to uncover actionable insights into human health and disease. 

A key milestone has been a nationwide analysis of Fitbit data from over 59,000 participants, revealing distinct temporal and geographic activity patterns and their associations with a broad spectrum of clinical outcomes. This ongoing collaboration also focuses on developing novel multimodal learning frameworks that jointly model diverse wearable signals and longitudinal clinical trajectories. 

To support the broader research community, ML4H contributes open-source pipelines for processing and modeling high-resolution wearable data. Together, these efforts reflect ML4H’s commitment to advancing precision health through open, interdisciplinary, and data-driven innovation.

Eric and Wendy Schmidt Center at the Ó³»­´«Ã½

ML4H collaborates with the , led by Caroline Uhler, on one of the center’s flagship projects focused on clinical trial emulation and drug repurposing. This ongoing effort leverages large-scale biobank and EHR resources — including the UK Biobank and NIH All of Us — to identify novel therapeutic opportunities grounded in real-world population data. The collaboration brings together EWSC’s expertise in causal inference with ML4H’s strengths in large-scale clinical machine learning. 

Together, the teams are developing and validating frameworks to infer causal relationships between existing drugs, biological phenotypes, and clinical outcomes — advancing the potential for repurposing approved therapies for new indications with scalable, data-driven methods.

Catalyst Fellowship Program at MIT linQ

In 2025, ML4H began a collaboration with the . The MIT Catalyst program brings together multidisciplinary teams to identify unmet healthcare needs and create technology-driven solutions. Fellows follow a structured process—from problem discovery and validation to solution design and research planning—aimed at developing projects ready for academic, clinical, or commercial implementation.

Through this partnership, ML4H members work closely with Catalyst Fellows to demonstrate how AI/ML modeling and biomedical data analysis can inform their projects and shape practical implementation strategies.