Summary
This peer-reviewed open-access study, published in the journal Digital Twin as part of the Digital Twin International Conference 2022 Collection, lays out BioTwin’s core approach: building an individual virtual twin by combining self-collected dried blood spot (DBS) sampling, untargeted LC-MS metabolomics, and a machine-learning workflow.
A cross-sectional study collected DBS samples from 277 volunteers over 30 months across Canada and the United States, with samples self-collected at home, mailed by standard post, and analyzed by UHPLC-MS. The work shows that metabolism is both dynamic within a person over time and distinctive between people: sample ownership was predicted with 80% accuracy when a user provided 5 samples and 92% accuracy with 10 samples. These results establish temporal variation and individuality as central features of metabolomic profiling.
Why it matters
This is the origin point of the BioTwin research programme. Every later result, the 1,257-participant identification preprint, the breast cancer detection study, and the physiological scores that feed the virtual twin, extends the proof of concept first reported here: that a single drop of dried blood, profiled by untargeted metabolomics, is enough to characterize an individual.
Authors
Manon Fradin, Louis-Philippe Noel, Gabriel Talbot-Lachance, Pierre Snell, Keven Voyer, Caroline Rheaume. Published in Digital Twin, 29 May 2024, 4:6 (DOI 10.12688/digitaltwin.17936.1), open access under CC BY. Most authors are employed by, and some hold shares in, BioTwin Inc.