Unsupervised integration of multimodal dataset identifies novel signatures of health and disease

Abstract

Modern medicine is rapidly moving towards a data-driven paradigm based on comprehensive multimodal health assessments. We collected 1,385 data features from diverse modalities, including metabolome, microbiome, genetics and advanced imaging, from 1,253 individuals and from a longitudinal validation cohort of 1,083 individuals. We utilized an ensemble of unsupervised machine learning techniques to identify multimodal biomarker signatures of health and disease risk. In particular, our method identified a set of cardiometabolic biomarkers that goes beyond standard clinical biomarkers, which were used to cluster individuals into distinct health profiles. Cluster membership was a better predictor for diabetes than established clinical biomarkers such as glucose, insulin resistance, and BMI. The novel biomarkers in the diabetes signature included 1-stearoyl-2-dihomo-linolenoyl-GPC and 1-(1-enyl-palmitoyl)-2-oleoyl-GPC. Another metabolite, cinnamoylglycine, was identified as a potential biomarker for both gut microbiome health and lean mass percentage. We also identified an early disease signature for hypertension, and individuals at-risk for a poor metabolic health outcome. We found novel associations between an uremic toxin, p-cresol sulfate, and the abundance of the microbiome genera Intestinimonas and an unclassified genus in the Erysipelotrichaceae family. Our methodology and results demonstrate the potential of multimodal data integration, from the identification of novel biomarker signatures to a data-driven stratification of individuals into disease subtypes and stages -- an essential step towards personalized, preventative health risk assessment.