Researchers Present New Multimodal Medical AI Model
The Apollo model integrates 25 billion patient records to improve clinical risk prediction and forecasting accuracy.
Updated on Oct. 26, 2026 in Cancer
Live Poll
Do you trust artificial intelligence to assist in medical diagnostic and treatment decisions?
Faisal Mahmood and his research team presented the Apollo multimodal temporal foundation model at the SIIM-CAIMI 2026 meeting. This new clinical system integrates diverse data types to provide rapid, probabilistic medical predictions for patients.
Why it matters
The system aims to make the entire patient record more computable by collapsing vast amounts of health data into a single vector. This approach allows researchers to synthesize complex medical histories for more accurate disease forecasting.
The Apollo model is trained on 25 billion records from 7.2 million patients across 28 modalities and 12 specialties. A study evaluated the model on 1.4 million patients, achieving AUC scores of 0.79 for breast cancer and 0.76 for lung cancer.
The players
Faisal Mahmood
Faisal Mahmood is a researcher who led the development of the Apollo model.
SIIM-CAIMI
SIIM-CAIMI is a professional organization that hosts meetings focused on imaging informatics and artificial intelligence in medicine.
The details
Apollo creates a single, temporally aligned representation of a patient's condition by integrating data from radiology, pathology, genomics, and clinical records. The system utilizes a model to pre-compute answers to clinical questions across medical notes to generate rapid outputs.
Timeline
April 2026: Study on Apollo clinical forecasting potential was published.
October 2026: Apollo model presented at the SIIM-CAIMI 2026 meeting.
The Big Picture
This development represents a shift toward consolidating fragmented patient data into unified foundation models. By outperforming established clinical risk scores including Fib-4 and PREVENT, the research marks a departure from reliance on single-modality prognostic tools.
The model could potentially reduce the number of placebo patients required for future clinical trials by approximately 25%. This shift aims to accelerate medical research and streamline the development of new treatments for patients.
The takeaway
The Apollo model transforms vast, siloed patient data into a single vector to improve predictive accuracy for serious illnesses. Researchers are now exploring how this computational power can be further applied to tools for 3D pathology and clinical trial optimization.
Further reading
Learn more about the latest research in the field of Cancer.
Live Poll
Do you trust artificial intelligence to assist in medical diagnostic and treatment decisions?
