Skip to navigation Skip to main content Skip to footer

Approved Research

Development and validation of multi-modal, weakly supervised machine learning (ML) models of phenotypic risk trained on high-dimensional, eigen-wavelet features

Principal Investigator: Dr Gordon Okimoto
Approved Research ID: 91100
Approval date: November 8th 2022

Lay summary

Data from UKB will be used to validate risk models developed by LifeDNA for approximately 150 human traits associated with an individual's health and wellness. The validated models will be aggregated in different ways using machine learning to define "super-traits" that describe an important aspect of an individual's health status in terms of multiple, interacting traits. The predicted risk for each super-trait will be displayed in LifeDNA's Dashboard of the Human Body (DHUB) to provide a comprehensive, real-time overview of an individual's health status at a specific point in time. DHUB will also provide near real-time feedback on how  modifiable risk factors for lifestyle, diet, environment, and physical activity interact with genetics, ancestry, and age to dynamically impact an individual's health and wellness over time. Successful completion of the project will be a step towards an online application that dynamically tracks an individual's health status based on hundreds of human traits while simultaneously assessing the emerging risk for certain disease conditions associated with specific patterns of phenome-wide risk.