Predicting age from binarized human oral microbial data combined with an ensemble of classifiers

Area of Science:

• Microbiology
• Gerontology
• Bioinformatics

Background:

• Human microbiome composition varies with age.
• Limited research exists on the oral microbiome's association with aging and its predictive potential.
• Age is a significant factor influencing microbial variation.

Purpose of the Study:

• To investigate the correlation between the oral microbiome and chronological age.
• To develop and validate a robust model for age prediction using oral microbiome data.
• To compare the performance of different data processing approaches for oral microbiome analysis.

Main Methods:

• Analyzed oral microbiome samples from 150 individuals aged 6-78 years.
• Utilized Permutational Multivariate Analysis of Variance (PERMANOVA) to identify age-related microbial variations.
• Developed an ensemble model using eXtreme Gradient Boosting (XGBoost) with 32 classifiers on binarized oral microbial data.

Main Results:

• Species richness and Chao1 index significantly increased with age.
• The ensemble model achieved a mean absolute error (MAE) of 7.20 years on an independent validation set.
• With increased sample size (2,550), the MAE decreased to 4.80 years, outperforming previous models.

Conclusions:

• Binarized oral microbial data combined with ensemble modeling is a promising approach for human age prediction.
• The developed XGBoost-based ensemble model demonstrates superior performance and robustness in age prediction.
• These findings provide a foundation for future research on microbiome-based age estimation.