AgeML: Age Modeling With Machine Learning.

Area of Science:

• Biomedical Informatics
• Computational Biology
• Gerontology

Background:

• Supervised machine learning enables age prediction from subject features, crucial for studying healthy and pathological aging across body systems.
• Current age modeling lacks standardized methodologies, hindering reproducibility and consistent reporting.

Purpose of the Study:

• To introduce AgeML, an open-source software designed to standardize and facilitate reproducible age prediction from tabular clinical data.
• To establish benchmarks for reporting in supervised age modeling tasks.

Main Methods:

• Developed AgeML, an open-source software package for age prediction using supervised machine learning on diverse tabular clinical datasets.
• Implemented functionalities for calculating age deltas, correlating them with various factors, and visualizing population-specific differences.
• Integrated classification of clinical populations based on age delta metrics.

Main Results:

• AgeML successfully reproduces previously published age prediction studies.
• The software identified novel associations between specific body organs and polygenic risk scores.
• Demonstrated the utility of AgeML in standardizing age modeling and enhancing reproducibility.

Conclusions:

• AgeML provides an accessible, standardized, and reproducible framework for supervised age modeling.
• The software facilitates deeper investigation into the biological underpinnings of aging and disease.
• AgeML simplifies complex age prediction tasks, promoting wider adoption and consistent research practices.