Electronic Health Record-Based Clinical Predictive Modeling to Match Care Coordination to Children with Complex Health Needs

Identification of high-risk children with complex health needs (CCHN) who may benefit most from care coordination to support intersecting medical and social needs is challenging yet essential to meet the health needs of children with high needs and their families. In this seminar, we will discuss the findings and lessons learned from development and implementation of the Pediatric Complex Care Integration (PCCI) program, in which we used machine learning clinical decision support and electronic health record (EHR) data-based predictive modeling to match complex care coordination services to CCHN within primary care at Duke.

We will highlight how lessons learned from this program advanced understanding of:
(a) how to implement child-specific clinical predictive modeling in real-world environments; and
(b) how to implement and evaluate the risk prediction data model in clinical practice using pragmatic methods.

We will also discuss opportunities to build on this project's blueprint and disseminate data-driven programs for use at the multi-site and population levels in alignment with value-based priorities and performance targets of health systems and payers.