Necessary storage Always on

Stores only your privacy choice.

Allow Google Analytics.

Learn more
Skip to content
BioTwin
EN
Language
TwinMe / Shop programs

BioTwin publishes medical research, clinical development, and wellness education. Clinical claims apply only where authorized. TwinMe wellness content is not medical advice, diagnosis, screening, treatment, or disease monitoring.

Publications Published 2 min

Metabolomic fingerprinting from dried blood spots enables individual identification across 1,257 participants at 94% user-level accuracy

Peer-reviewed open-access study in Digital Twin (Taylor & Francis) by Hauguel, Anctil, and Noel, showing that metabolomic profiles from dried blood spots are stable enough to identify individuals across 18,288 samples and 134 analytical batches.

Pierrick Hauguel, Nicolas Anctil, Louis-Philippe Noel ·
biometricsmetabolomicsdried blood spotspublished

Why it matters

Establishes the methodological foundation for personal-baseline interpretation of longitudinal biomarker data, the core mechanic behind BioTwin's virtual twin model. The same-lab DBS LC-MS protocol and the GroupKFold batch-aware validation standard introduced here underpin all of BioTwin's downstream disease classification work.

View publication →

Summary

A 1,257-participant cohort with 18,288 dried blood spot (DBS) samples collected over 134 analytical batches and 15 months is used to test whether untargeted LC-MS metabolomic profiles carry enough individual-level signal to identify a participant from a single fingerprick. After batch-aware normalization, supervised feature selection, biological signal filtering, and user-level majority voting across DBS cards, the model reaches 94.1% user-level accuracy and 85.5% sample-level accuracy under 10-fold GroupKFold (group = batch). On a held-out future-batch set of 17 batches, the model reaches 96.1% user-level and 92.6% sample-level accuracy across 1,134 classes, against a chance baseline of 0.088%.

A second contribution of the paper is methodological: the authors show that naive random splitting inflates accuracy because 92.8% of test samples share their (user, batch) pair with the training set. Group-aware splitting is required to measure real generalization.

Why it matters

Most biomarker science still treats lab values as one-shot snapshots. This study lays out the case, with data, for treating biology as a trajectory and for evaluating change against an individual’s own baseline. It also frames the protocol and validation discipline that the rest of BioTwin’s research programme is built on.

The outcome studied is individual identification. Its accuracy does not establish the clinical validity of a health score, disease prediction or treatment recommendation.

Authors

Pierrick Hauguel, Nicolas Anctil, Louis-Philippe Noel, Research and Development Department, BioTwin Inc. Published in Digital Twin (Taylor & Francis) on 21 August 2026, DOI 10.1080/27525783.2026.2706347, open access under CC BY. Received 16 April 2026, accepted 11 July 2026. All authors are employees and shareholders of BioTwin Inc. PCT patent pending.

Important: This article may discuss BioTwin research, medical vision, regulated clinical pathways, or TwinMe wellness education. TwinMe wellness outputs are not medical or laboratory tests. BioTwin clinical outputs are available only where authorized and through licensed healthcare professionals.