Related Experiment Videos
A Machine Learning-Based Model to Predict Overactive Bladder Risk Among US Women: Evidence From the National Health
Guoqiang Huang1, Shuangquan Lin1
1Department of Urology, The Second Affiliated Hospital of Nanchang University, 1 Mingde Road, Nanchang, Jiangxi Province, 330200, China, 86 15727561896.
JMIR Medical Informatics
|August 12, 2026
Summary
Machine learning accurately predicts overactive bladder (OAB) risk in women using reproductive and sociodemographic factors. Key predictors include age, BMI, and number of vaginal deliveries, aiding early detection.
Area of Science:
- Women's health research
- Medical informatics
- Predictive analytics
Background:
- Overactive bladder (OAB) is common, especially in women, causing urgency, frequency, and nocturia.
- Current OAB risk prediction is limited, often excluding crucial reproductive history.
- Machine learning (ML) can enhance OAB prediction using large datasets like NHANES.
Purpose of the Study:
- Develop and validate an ML model for predicting OAB risk in women.
- Incorporate reproductive and sociodemographic factors into the OAB prediction model.
- Identify key OAB predictors using interpretable ML methods.
Main Methods:
- Analyzed 7884 participants from NHANES (2011-2018) in a retrospective observational study.
- Utilized LASSO regression for variable selection and random forest for model construction.
- Employed SHAP and RCS for model interpretation and dose-response analysis.
Main Results:
- The random forest model showed moderate predictive capability (AUROC 0.6999 in test set).
- Top predictors identified: age, BMI, and number of vaginal deliveries.
- Positive associations found between OAB risk and age, BMI, earlier menarche, and more vaginal deliveries.
Conclusions:
- ML models integrated with SHAP offer robust OAB risk prediction.
- This approach facilitates early identification and improved clinical management of OAB.
- Reproductive and sociodemographic factors are crucial for accurate OAB risk assessment.
