Enabling Cost-Effective Population Health Monitoring using Artificial Intelligence and Machine Learning

Existing solutions for monitoring and modeling population health, such as curating and linking data from electronic health records (EHR) or using labor intensive health surveys, are very costly, have limited spatiotemporal coverage and scale, which limits their suitability in supporting the needs of healthcare and health policy making in terms of cost, time, and accuracy. Another limitation of these solutions is that they mostly target monitoring conditions that have already developed rather than offer insights of invisible trends that can be used to help design policies for their early detection and prevention.

To address these challenges, this project, jointly conducted with the Centre for Intelligent Healthcare, Coventry University, investigates a cost-effective approach called Compressive Population Health (CPH), where a subset of a given geographical area is selected in terms of regions within the area for data collection in the traditional way, while leveraging inherent spatial correlations of neighboring regions to perform data inference for the rest of the area. By alternating selected regions longitudinally, this approach can validate and correct previously assessed spatial correlations. We are validating our approach by conducting in-depth studies based on spatiotemporal morbidity rates of chronic diseases in several areas, including a published study of more than 500 regions around London, UK, for over ten years. Preliminary results can be read in the ACM Transactions on Computing for Healthcare (check Publications, 2021, Digital Health).

We are also working on applying our CPH approach to solve the missing disease prevalence records problem, which is a common place in population health datasets. Missing records challenge the utility of the health data and hinders reliable analysis and understanding. To tackle this problem, we are investigating the use of a deep-learning-based version of CPH, to infer and recover (to complete) the missing prevalence rate entries of multiple chronic diseases. Results can be read in the Proceedings of The Web Conference (previously known as the WWW Conference) which can be found in the Publications section, 2021, Digital health.