Our research addresses three interconnected aims:
1. We developed a deep-learning biological age clock using only 7 blood-based standard clinical biomarkers from a South-East Asian hospital cohort, submitted to Scientific Reports in December 2024. The feedback was very positive, and we were encouraged to validate our findings in an independent, ethnically distinct cohort to assess generalizability. One reviewer requested additional validation to demonstrate further robustness and cross-ethnic applicability of our aging clock.
2. In parallel, we plan to reimplement and retrain this aging clock using the same 7 biomarkers (HDL, triglycerides, creatinine, HbA1c, alanine aminotransferase and albumin and sex) in the UK Biobank to explore population-specific calibration and enhance predictive accuracy for European cohorts. We will compare the clock’s predicted biological age with participants’ chronological age and assess its associations with diverse meta-data, including disease status, mortality, and bio-psycho-social variables. We will stratify individuals within each age cohort into quartiles based on relative biological age and determine healthy ranges for all available biomarkers.
3. Apart from the validation and the aforementioned project(s), we plan to establish the feasibility of building a biological age clock on the intensity of physical activity determined by accelerometer data. This approach will enable us to investigate the relationships among physical activity intensity, mortality, and the predicted biological ages from our newly developed clock. Furthermore, we will compare these associations with other available metadata to identify potential confounders or variables of interest.
This integrated approach will provide comprehensive insights into aging across different populations and modalities and enhance personalized preventive healthcare.