Type 1 diabetes (T1D) is a lifelong chronic disease affecting more than 1.7 million Americans, including over 300,000 children and adolescents. To better prevent and treat this disease, it’s important to fully understand how and why it develops.
The new Multimodal AI for Type 1 Diabetes (MAI-T1D) project — funded by the National Institutes of Health — will use the latest advances in artificial intelligence (AI) alongside a wide range of health data with the goal of accelerating discoveries to better treat and prevent T1D.
MAI-T1D is a collaboration between scientists from leading institutions, including the University of Michigan; Vanderbilt University Medical Center; University of California, Los Angeles; University of South Florida and Weill Cornell Medicine. The group will build AI models to examine information like genetics, proteins and changes in individual cells to better map how T1D develops — from the signs of immune system changes through to symptoms and diagnosis.
The aim is to uncover what causes the immune system to attack insulin-producing cells, understand how these cells decline, and find ways to prevent or delay T1D — even before it causes symptoms.
To train and test these AI models, the team will use large, detailed datasets from the Human Pancreas Analysis Program (HPAP), which studies pancreas samples from healthy people and people with T1D at various stages, with a commitment to scientifically rigorous and responsible use of data. The researchers will also use data from TEDDY (The Environmental Determinants of Diabetes in the Young), an international study that follows children who are genetically at risk for T1D, tracking their health and environmental exposures over time.
Led by Jie Liu, PhD (University of Michigan), the MAI-T1D project brings together more than 25 scientists with experience in AI, genetics and diabetes research. Their ultimate goal is to make it possible to predict and prevent the disease more precisely, improving lives for those with or at risk for T1D.
Key investigators include Shuibing Chen, PhD (Weill Cornell Medicine; single-cell and spatial multiomics); Marcela Brissova, PhD (VUMC; islet biology, HPAP leadership); Kenneth Young, PhD (University of South Florida; AI and bioinformatics for TEDDY, TrialNet, Rare and Atypical Diabetes Network); Kai-Wei Chang, PhD, and Wei Wang, PhD (UCLA; multimodal AI and foundation models); and Stephen Parker, PhD (University of Michigan; diabetes genetics, integrative multiomics).
