The Challenge
Mining unstructured clinical notes for insights and automating documentation; predicting patient outcomes from text data.
The AI Solution
Developed an in-house large clinical language model trained on years of NYU clinical notes, plus predictive analytics to identify at-risk patients.
Type of AI Used: Generative AI (Custom LLM) and Predictive Modeling
Results & Impact
NYU's proprietary clinical AI model demonstrated up to 92% accuracy on certain outcome predictions. Used to flag patients at risk based on doctor notes.
