Integrative adaptive indexes from noisy routine haematological markers can predict and discriminate health status and biological age

For more than two decades, advances in personalised medicine and precision healthcare have largely been based on genomics and other omics data. These strategies aim to tailor interventions to individual patient profiles, promising greater treatment efficacy and more efficient allocation of healthcare resources. Here, we show that widely collected common haematological markers can reliably predict and discriminate individual chronological age and health status from even noisy sources. Our analysis includes synthetic and real retrospective patient data, including medically relevant and extreme cases. We combine fully explainable risk assessment scores with machine intelligence to focus on clinically significant patterns and characteristics without functioning as a “black box” model and allowing interpretation and control tested on the US CDC NHANES database (100 000 participants) and validated with the UK Biobank (500 000 participants). Despite the noisy nature of these databases due to self-reporting and sparse data, the results remained relevant and statistically significant. Unlike current biological ageing indicators, this approach may offer rapid, and scalable implementations for personalised, precision and predictive approaches to healthcare and medicine without or before requiring other specialised, uncommon, or costly tests. • A single-value numerical risk assessment index from routine CBC markers is introduced. • The index can discriminate health even in noisy data for triaging and stratification purposes. • The index can predict health status and biological age derived from blood counts related to inflammaging. • The index was tested and validated across 600K+ subjects from US CDC NHANES and UK Biobank datasets. • In combination with a neuro-symbolic AI approach, it enables explainable and scalable health stratification.