Researchers have developed new neural network models that predict human biological age based on blood biochemical markers and gut microbiome species, according to a March 23 announcement. The study was published in Volume 18 of Aging-US on March 12.
The research is led by Anastasia A. Kobelyatskaya from the Russian Clinical Research Center for Gerontology, Pirogov Russian National Research Medical University, and the Institute of Biology of Aging and Healthy Longevity Medicine with Preventive Medicine Clinic, Petrovsky Russian Research Centre of Surgery. Alexey Moskalev served as corresponding author.
The team created two types of models: a gender-specific biochemical model using seven routine clinical markers such as cystatin-C, IGF-1, DHEAS (plus sex-specific sets), and a microbiota model analyzing 45 gut bacterial species measured through full-length 16S sequencing. Both models were trained and tested on data from 637 individuals, achieving mean absolute errors around six years and R² values above 0.8.
To make these models more interpretable for clinicians and researchers, the authors applied SHapley Additive exPlanations (SHAP). This method helps show how individual predictors—such as DHEAS or specific bacterial species like Blautia obeum—influence the predicted biological age for each person. The biochemical clock used only seven clinically accessible markers to ease its adoption in healthcare settings while the microbiota clock highlighted key taxa whose abundance correlates with predicted microbiotic age.
"As the proposed models possess both global and local explainability, they hold future potential for application in monitoring the effectiveness of various interventions in clinical trials," said the authors.
However, limitations remain: the study cohort was limited to a Caucasian population and implementing the microbiota model requires significant sequencing resources. The authors call for external validation in larger populations with greater ethnic diversity, prospective testing to connect predictions with health outcomes, and use of these explainable tools to monitor responses in intervention studies where changes in predicted biological age could serve as early signals.
