Related Experiment Video
Updated: Aug 26, 2025
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
7.6K
A machine learning-based data mining in medical examination data: a biological features-based biological age
Qing Yang1, Sunan Gao2, Junfen Lin1
1Zhejiang Provincial Center for Disease Control and Prevention, Hangzhou, 310051, China.
BMC Bioinformatics
|October 3, 2022
Summary
This study developed a novel machine learning-based biological age (ML-BA) model for the Chinese population, improving aging assessment accuracy and stability. The new composite ML-BA (STK-BA) effectively addresses overfitting and enhances health status predictions.
Area of Science:
- Gerontology
- Biostatistics
- Computational Biology
Background:
- Biological age (BA) is a superior aging metric compared to chronological age (CA).
- Existing BA models face challenges with incomplete medical data and population-specific applicability.
- Machine learning-based BA (ML-BA) development requires addressing model overfitting for stable health associations.
Purpose of the Study:
- To develop and validate a robust ML-BA for the Chinese population.
- To evaluate data imputation methods and assess ML-BA overfitting.
- To create a composite ML-BA that enhances stability and health correlation.
Main Methods:
- Evaluated various missing data interpolation techniques, identifying round-robin linear regression as optimal.
- Constructed 14 ML-BAs using diverse biomarkers from Chinese adults aged 45-90.
- Developed a Stacking-based composite ML-BA (STK-BA) to mitigate overfitting.
Main Results:
- AutoEncoder demonstrated the highest interpolation stability.
- The proposed STK-BA model overcame overfitting issues.
- STK-BA showed strong correlations with chronological age (r=0.66), health indicators, disease counts, and specific diseases.
Conclusions:
- An improved aging measurement method for Chinese middle-aged and elderly populations was developed.
- The STK-BA model offers stable aging characteristic capture beyond chronological age.
- This research highlights the potential of ML in advancing aging research and applications.
