MIT Researchers Develop AI Model to Detect Pancreatic Cancer Earlier
• MIT researchers developed 2 machine learning models, PRISM, to detect pancreatic cancer earlier than current methods • PRISM was trained on over 5 million patients' health records, making it more diverse than other models focused on specific regions • It can identify pancreatic cancer 35% of the time, compared to 10% detection rate of current standards • The goal is to diagnose pancreatic cancer earlier - 80% of cases are caught in late stages currently • PRISM analyzes patient data like demographics, diagnoses, medications, lab results to predict cancer probability