Home > Events > 2014 Seminars & Colloquia > NICOLETA SERBAN - Georgia Institute of Technology

NICOLETA SERBAN - Georgia Institute of Technology

Main Content

Profiling Utilization in the Medicaid System from Large, Highly Sensitive Patient-level Claims Data
30 October 2014 from 4:00 PM to 5:00 PM
201 Thomas Bldg.
Add event to calendar

Health Analytics has become ubiquitous in the pursuit of understanding and managing the complexity of the healthcare system. The significant efforts towards health(care) data acquisition have prompted a high demand for adaptive and scalable statistical methods, providing the means for proactively transforming healthcare. The research presented in this seminar is an illustration of bridging fundamental and computational modeling with health services research as a means of translating healthcare data into knowledge and decision making.

We use the Medicaid Analytic Extract (MAX) claims data acquired from the Centers for Medicare and Medicaid Services (CMS) to demonstrate how to map patient-level claims information to underlying utilization profiles. The objective of this study is to investigate a statistical and graphical approach for uncovering patterns in healthcare utilization within the Medicaid system. We assume a Markov renewal process (MRP) to model the patient-level utilization sequences and profile them into distinct clusters. We use the Bayesian information criterion (BIC) to determine the optimal number of clusters and the Kullback-Leibler (KL) distance to determine the optimal clustering region. We seek to provide inference on patient care profiles when viewed in a longitudinal manner. We pilot this study for the set of children with an asthma diagnosis over five years of Medicaid claims data and two neighboring states, Georgia and North Carolina.

This research is collaborative with two Ph.D. students at Georgia Institute of Technology, Ross Hilton and Yuchen Richard Zheng. The project has been supported by seed funds from Children's Healthcare of Atlanta and the Institute of People and Technology.