Background: A digital health disease management company partnered with the Center for Health Information and Decision Systems to gain insights into their patient population and the success of their program.
Goal: The goal of this project is to identify patient characteristics and program features that contribute to patient success.
Objectives: 1. Combine all data at the patient level to prepare for machine learning analysis. 2. Identify measures of patient success. 3. Identify trends in patient characteristics that point toward patient success. 4. Identify program characteristics that lead to patient success.
Approach: Data is coming into the company's system from many different sources. We combined and mined data on insurance claims, initial patient health survey (similar to PHQ-9), device reading data streaming into the system several times per day, notes on health coaching sessions, notes on outreaches triggered by the device reading system and demographic data. We continue to work on preparing the data for time series analysis.
Results: Right now, we are building the dataset and running some analytics. We do not have conclusive results at this time, but we are hoping the time series analysis will help us. Also, the company is enrolling more patients and we hope a larger dataset can run through our scripts and point to patient indicators.
Importance to public health: The patients are Medicaid patients and if programs like this can help Medicaid patients manage their disease conditions, the hope is that they can be applied to other patients as well.