Multimodal Generative AI for Precision Health
Hoifung Poon, PhD
The dream of precision health is to develop a continuous learning health system where new health information is instantly incorporated to optimize care delivery and accelerate biomedical discovery. Multimodal generative AI has the potential to drastically accelerate progress toward precision health by supercharging the structuring of health data and scaling population-level insight generation. In this article, we will highlight several prominent growth areas in this exciting frontier:
Application: Prior health AI applications often centered around diagnostics, but there are many low-risk yet high-value scenarios across the entire health system that are ripe for impact.
Evaluation: Real-world use cases are often under-represented in existing health AI benchmarks; scaling realistic benchmark creation and evaluation is of increasing urgency.
Modeling: Unlike standard contrastive learning, multimodal generative AI can benefit from gravitating in text as the “interlingua” of all modalities, given the vast amount of human knowledge captured in state-of-the-art large language models.