Building gender-specific sexually transmitted infection risk prediction models using CatBoost algorithm and NHANES...

Area of Science:

• Public Health
• Machine Learning
• Epidemiology

Background:

• Sexually transmitted infections (STIs) pose a significant global health challenge with increasing incidence and disease burden.
• Trends from 2010-2019 show rising age-standardized rates for syphilis, chlamydia, trichomoniasis, and genital herpes.
• Machine learning (ML) offers powerful tools for predicting disease outbreaks and individual risk.

Purpose of the Study:

• To develop and validate male- and female-specific STI risk prediction models using the CatBoost algorithm.
• To analyze individual STI prediction and overall STI risk.
• To identify key demographic and behavioral predictors of STIs in different populations.

Main Methods:

• Utilized data from the National Health and Nutrition Examination Survey (NHANES) on 12,053 participants (ages 18-59).
• Employed the CatBoost algorithm for prediction, with ADASYN to handle data imbalance and SHAP for feature importance.
• Evaluated 15 ML algorithms before selecting CatBoost for its performance.

Main Results:

• CatBoost achieved high AUC values for predicting various STIs in males (e.g., 0.9995 for chlamydia) and females (e.g., 1 for gonorrhea).
• Significant predictors for male STIs included new partners/year, condomless sex/year, and lifetime male partners.
• Key female STI predictors were anal sex with a man, age, and lifetime female partners.

Conclusions:

• The CatBoost classifier effectively predicts STI risks in both male and female populations.
• SHAP analysis identified crucial demographic and behavioral predictors for STIs.
• Findings can inform targeted prevention strategies to reduce the public health impact of STIs.