APOLLO is a multimodal temporal foundation model trained on 25.2 billion medical events from 7.2 million patients across 33 years of Mass General Brigham records, integrating 28 medical modalities into unified ?virtual patient representations.? Across 322 downstream tasks, the model showed strong performance in disease prediction, treatment-response stratification, hospital operations, and patient retrieval, supporting the idea of more computable, data-driven medicine.
More...