Multimodal Omics and Wearable Sensor Data Fusion for Continuous Cardiometabolic Health Assessment

Abstract

Background: Diabetes, hypertension, and cardiovascular diseases are cardiometabolic diseases that are the major causes of morbidity and mortality globally. Conventional approaches to health monitoring are not very sensitive to the dynamic nature of these states, which is why continuous, real-time health measurements are required. Objectives: The proposed research endeavors to incorporate multimodal omics data (genomic, proteomic, and metabolomic) with wearable sensor data for continuous cardiometabolic health monitoring and to construct a predictive model to evaluate cardiometabolic risk using a combination of these varied data sets. Materials and Methods: The genomic, proteomic, and metabolomic data were obtained among 150 participants with different cardiometabolic risks, who were wearing smartwatches and continuous glucose monitors to measure their heart rate, blood pressure, and glucose level. Preprocessing was done on the data, and machine learning models, including deep learning models, were employed to combine data and predict risks. Results: The predictive model had an accuracy of 85% and an AUC-ROC of 0.92, which is significantly higher than that of traditional clinical measures (70% accuracy; AUC-ROC = 0.76). This model showed great predictive capacity regarding the integration of omics with sensor data to better predict cardiometabolic risk. Conclusion: The paper provides evidence of the opportunities in the combination of omics data and wearable sensors to enable continuous and personalized cardiometabolic health monitoring, achieving a path to precision medicine and preventive disease treatment.