Personalized-Context-Aware Age Gap: A New Multi-Omics Measurement Based on Age-Enhanced Model AOE-Net for Aging...

Area of Science:

• Gerontology and computational biology.
• Development of novel biomarkers for aging.
• Multi-omics data analysis for health prediction.

Background:

• Aging is a global health challenge, increasing disease risk.
• Traditional age gap (AG) metrics have limitations in capturing stratified aging patterns.
• Existing AG calculations may lead to biased interpretations of aging acceleration.

Purpose of the Study:

• To introduce Personalized-context-Aware Age Gap (PAAG), a novel metric for estimating aging acceleration.
• To develop and validate AOE-Net (Age Order Enhanced Network), a pre-training model for PAAG.
• To improve the prediction of clinical outcomes for age-related diseases.

Main Methods:

• Developed AOE-Net using age-order enhanced contrastive learning on multi-omics data from healthy individuals.
• Learned latent representations to capture biological deviation in aging trajectories.
• Fine-tuned AOE-Net to generate PAAG and compared its predictive performance against conventional aging clocks.

Main Results:

• PAAG significantly outperformed traditional AG in predicting clinical outcomes across pan-cancer, atherosclerosis, and osteoporosis.
• Validated PAAG's superior predictive power across diverse age-related diseases and phenotypes.
• Identified immune-response pathways as key molecular drivers linking accelerated aging and disease via PAAG analysis.

Conclusions:

• PAAG is a robust, context-aware metric for assessing aging acceleration and predicting clinical outcomes.
• AOE-Net serves as an effective pre-training model for aging research and PAAG evaluation.
• PAAG offers a stable indicator for clinical assessment of age-related diseases and their underlying mechanisms.