AI-Based Solutions for Enhancing Health Data Integration: Leveraging Machine Learning to Combine Electronic Health Records, Genomic Data, and Wearable Sensors

Abstract

Health data integration is becoming increasingly important to create a comprehensive picture of patient health status and needs, especially as these extend beyond the clinical setting. There is agreement among health data regulators that integrating at least three domains is relevant if the aim is clinical utility. Mapping the genomes of patients and controlling the transcriptome and proteome profiles have begun to be important when creating individualized drug treatment plans. Health data collected using wearable sensors are also informative about patient behavior and response to non-genomic variables. Wearables provide information about activities of daily living, sleep, and some body functions. On the one hand, the problem with these independent data sources that contribute data from these domains is that they are all independent, causing us to lose the bigger picture. Therefore, it is important to use AI and machine learning techniques that can integrate and align such a majority of data from both non-invasive and invasive sources. This paper briefly discusses potential integration techniques of electronic health records, genetics, and wearable data and highlights some findings.