ExplaiNAble BioLogical Age (ENABL Age): an artificial intelligence framework for interpretable biological age.

Area of Science:

• Computational biology
• Biotechnology
• Genomics

Background:

• Biological age is a key health indicator, but current methods for assessing it often lack accuracy or interpretability.
• Existing age clocks present a trade-off between precision and the ability to understand individual health factors.

Purpose of the Study:

• To introduce ExplaiNAble BioLogical Age (ENABL Age), a novel computational framework designed for accurate biological age estimation.
• To integrate machine learning with explainable artificial intelligence (XAI) for individualized health insights and interpretable age predictions.

Main Methods:

• Developed ENABL Age by predicting age-related outcomes and rescaling them to estimate biological age using UK Biobank and NHANES data.
• Adapted XAI methods to identify individual risk factors contributing to biological age, creating blood-based (ENABL Age-L) and questionnaire-based (ENABL Age-Q) versions.
• Validated the distinct aging mechanisms captured by ENABL Age clocks through genome-wide association studies (GWAS).

Main Results:

• ENABL Age clocks showed significant correlation with chronological age (r=0.7867 to 0.7126) and effectively distinguished healthy from unhealthy individuals.
• The clocks demonstrated superior mortality prediction (AUC 0.8179-0.9107) compared to existing methods.
• Individualized explanations revealed key aging characteristics, and GWAS confirmed that each clock captures unique aging mechanisms.

Conclusions:

• ENABL Age represents a significant advancement in applying XAI to biological age clocks, offering enhanced interpretability.
• The framework has substantial potential for clinical settings, aiding in understanding aging complexities and supporting informed decision-making.