Deep Learning for Medical Chart Classification
Knowing the evidence of a disease as a current condition for a member is critical for areas like risk adjustment, planning interventions, better personalization, etc that might not always be evident from claims data alone. A member’s present health condition can be in the EHR (Electronic Health Record) in the form of doctors notes that may not be in claims data for a time period. Diseases mentioned in EHR can be in different contexts like family history, past conditions, etc. The talk will present our work on attention based deep learning approaches for classifying a chart for a health status in the context of a current condition
Lakshmi Manohar Akella is a Data Science/AI Director in the Enterprise Data Science team with Anthem. He has over 14 years of experience in conducting applied research and developing Machine Learning approaches for a variety of problems in e-commerce, health care and pharmaceutical industries. He leads ML projects end to end from conception of an algorithmic idea to production of real-time machine learning systems. Currently he works on Deep Learning methods for clinical event prediction and disease progression modeling.