Improving Pelvic Floor Muscle Training with AI: A Novel Quality Assessment System for Pelvic Floor Dysfunction

Area of Science:

• Urology
• Biomedical Engineering
• Artificial Intelligence in Healthcare

Background:

• Pelvic floor muscle (PFM) training is the primary treatment for urinary incontinence.
• Many women face challenges with correct PFM contraction technique and understanding.
• Accurate assessment of PFM contraction quality is crucial for effective rehabilitation.

Purpose of the Study:

• To develop and validate a novel AI-based system for assessing PFM contraction quality.
• To enable autonomous monitoring and improvement of PFM contractions for women undergoing rehabilitation.
• To address the limitations of current methods in evaluating PFM training effectiveness.

Main Methods:

• Development of a PFM contraction detector using a convolutional neural network.
• Implementation of a maximal PFM contraction performance classifier with a custom feature extractor and random forest.
• Training and testing AI algorithms using vaginal dynamometry data, supplemented by expert physiotherapist assessments.
• Validation of algorithms against the modified Oxford scale for contraction strength.

Main Results:

• The PFM contraction detector achieved 97% and 100% accuracy on independent test datasets.
• The contraction performance classifier accurately predicted strength ratings within ±1 scale point with 97% accuracy.
• The system demonstrated the ability to extract clinically relevant features automatically with acceptable error.

Conclusions:

• The developed AI system offers a reliable and accurate method for assessing PFM contraction quality.
• This technology has significant potential to enhance PFM training and rehabilitation programs.
• Women can utilize this system for autonomous self-monitoring and improvement of their PFM exercises.