Blood biochemical and gut microbiotic neural network models forecasting human biological age

Biological age reflects the current state of the body, considering the aspects of lifestyle, environment, and hereditary component. Currently there is no universal formula for determining it, but there are markers that can be used to calculate it. This study aims to develop and compare two models for calculating biological age based on laboratory blood tests and composition of gut microbiota. The biochemical model of biological age uses 7 indicators and is gender-specific (general – cystatin-C, IGF-1, DHEAS, only for females – homocysteine, urea, glucose, zonulin, only for males – HbA1c, NT-proBNP, free testosterone, hs-CRP). The microbial model requires the input of percentages of 45 bacterial species as indicators of the gut microbiota. Both methods demonstrate high predictive accuracy (MAE ~ 6 years, R2 > 0.8) and the degree of agreement of assessments both with each other and with PhenoAge (correlation > 0.89). For enhanced interpretability of the models, we applied the SHAP explanation algorithm, which made which allowed us to evaluate the contribution of each predictor to the final assessment of the biological age.