Experts at UC Berkeley and UC San Francisco are taking a new approach to the treatment of diabetes and metabolic health by using an open source platform.
The Agile Metabolic Health project is being led by Ida Sim, M.D., Ph.D., co-director of the two campus’ Joint Program in Computational Precision Health (CPH), and Fernando Pérez, Ph.D., incoming faculty director of the Berkeley Institute for Data Science (BIDS).
More than 38 million Americans have diabetes, with African Americans and other people of color disproportionately affected. The inaugural effort of Open Platforms for Powering Science and Society initiative at the UC Berkeley College of Computing, Data Science, and Society, Agile Metabolic Health will pilot a new approach to the development of algorithms and clinical decision support using data from wearable devices, such as blood pressure cuffs and continuous glucose monitoring, to be deployed first at UCSF’s General Internal Medicine, Diabetes, and Weight Management clinics.
Data from wearables and algorithms used to interpret them are generally closely guarded by device makers and siloed by disease, the Agile Metabolic Health team notes. By contrast, Agile Metabolic Health will be powered by JupyterHealth – an open platform that facilitates the secure collection of health data from sensors and medical records, and supports the development, sharing and testing of digital biomarker algorithms for clinical use.
JupyterHealth will use open source software, shareable notebooks and code to support reproducibility, and vendor-agnostic infrastructure – to enable new research collaborations for public benefit and transform health algorithm development.
The platform will also provide patients and physicians new ways to personalize diabetes management, providing real-time insights into blood sugar fluctuations. Initially focused on metabolic disorders, JupyterHealth aims to expand to enable data collection, development, and testing of algorithms across various chronic diseases, which impact nearly 40 percent of the U.S. population.
Team members include developers from 2i2c, the organization that manages the design, configuration and linkage between Jupyter datahubs now used by more than 10 million researchers and educators worldwide; the Commons Project Foundation, a global tech nonprofit pioneering approaches to patient-centered, privacy-protecting management of health information; and the Duke University Big Ideas Lab, specializing in the analysis of multimodal data and development of biomarkers to predict health risk.
