Detection of the most influential variables for preventing postpartum urinary incontinence using machine learning techniques
This work addresses a common health issue for postnatal women by providing data-driven insights for prevention, though it is incremental as it builds on prior variable identification with new analysis.
The study tackled predicting postpartum urinary incontinence (PUI) by identifying influential variables using machine learning on data from 93 pregnant women, finding that models using extrinsic variables achieved accuracies of 70% to 93% for different PUI outcomes.
Background: Postpartum urinary incontinence (PUI) is a common issue among postnatal women. Previous studies identified potential related variables, but lacked analysis on certain intrinsic and extrinsic patient variables during pregnancy. Objective: The study aims to evaluate the most influential variables in PUI using machine learning, focusing on intrinsic, extrinsic, and combined variable groups. Methods: Data from 93 pregnant women were analyzed using machine learning and oversampling techniques. Four key variables were predicted: occurrence, frequency, intensity of urinary incontinence, and stress urinary incontinence. Results: Models using extrinsic variables were most accurate, with 70% accuracy for urinary incontinence, 77% for frequency, 71% for intensity, and 93% for stress urinary incontinence. Conclusions: The study highlights extrinsic variables as significant predictors of PUI issues. This suggests that PUI prevention might be achievable through healthy habits during pregnancy, although further research is needed for confirmation.