Interventional Population Health
Joel Saltz, Chair, describes this work:
"We are developing an integrated data-gathering, data-integration, predictive-modeling and information-dissemination infrastructure. This integrated system can: 1) reduce the risk of undesirable events like potentially avoidable emergency department visits and hospital-acquired infections; and 2) improve patient quality indicators such as measures of diabetic control and psychiatric disease management.
BMI is developing 360-degree surveillance and prediction methods designed to catpure and interate multiple complementary sources of information and to use this integrated information to make targeted, actionable predictions. The information we leverage is derived from:
- Electronic health record's discrete data elements
- Information extracted from text via Natural Language Processing
- Wearable sensors
- Patient-reported and clinician-reported information obtained from secure smartphone web apps
- Public data sources (e.g., American Community Survey and U.S. Census).
We use a variety of machine learning methods to integrate data and make predicitons, which are translated into actionable reminders by integration into the electronic health record and through development of FIHR-based clinician- and patient-facing mobile web apps."