Healthome Polygon Framework: A Comprehensive and Multidimensional Health Quantification Framework Using Artificial Intelligence and Multiomics Data

BACKGROUND: Quantifying human health and disease necessitates a transformative framework capable of integrating diverse biomedical data in a standardized manner. Such a system could enable precise preemptive health and disease prevention through the application of artificial intelligence. OBJECTIVES: We propose a novel health quantification framework using Pseudo Super Healthy (PSH) subjects—virtual optimal health conditions, and the Multi-Omics Health Index (MOHI) which consolidates health scores across eight “omes”: Physiome, Metabolome, Vasculome, Inflammatome, Immunome, Psycholome, Transcriptome, and Epigenome. METHODS: PSH subjects were identified using algorithmic or clinical domain knowledge-based methods in the first five omes, utilizing data from the Korea National Health and Nutrition Examination Survey and the Korean Genome Project (KGP). Individual health status in these omes was quantified using Euclidean distances (ED) from PSH subjects. For the last three omes, machine learning (ML) models trained on KGP data predicted individual health probabilities. RESULTS: PSH subjects were stratified by age and sex. In two available case studies—a male in his mid-50s and a female in her mid-30s—their MOHI scores were 30 and 49, respectively. ED-based health indicators correlated significantly with the Number of Abnormal Traits (NATs) in Physiome (Pearson’s r = 0.62 for males, 0.70 for females, _P_ < 0.001). Three ML models used to quantify health for the Psycholome, Transcriptome, and Epigenome achieved area under the curve values of 0.79, 0.85, and 0.96, respectively. CONCLUSIONS: Individual health can be represented by a radar chart (Healthome Polygon) derived from PSH subjects and ML models, resulting in MOHI scores that serve as a personalized health quantification metric.