HALICON 2017

A surge in health care data, including electronic health records and medical sensor data, is providing new opportunities to uncover data-driven insights and patterns of disease using machine learning techniques. Sanjay Joshi and Karen Hayrapetyan will present a summary of AI in Health and two separate use cases that could reduce cost and increase care.

The first use case is predicting diabetes hospital readmissions using H2O's gradient boosted trees using the diabetes dataset from 130 US hospitals For the second use case we present a geo-spatial forecasting system for seasonal communicable disease that has the potential to aid public health planning and help reduce the impact of outbreaks, specifically for time-series forecasts of Influenza-like illness (ILI) encounters at county and zip-code levels with increased performance compared to conventional systems that predict ILI incidents at isolated spatial scales