Related Experiment Video
Updated: Sep 19, 2025
Quantification of Levator Ani Hiatus Enlargement by Magnetic Resonance Imaging in Males and Females with Pelvic Organ Prolapse
Published on: April 17, 2019
AI-based pelvic floor surface electromyography reference ranges and high-precision pelvic floor dysfunction
Juan Chen1, Jiahui Yao2, Wei Chen3
1Department of Obstetrics and Gynecology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College; National Clinical Research Center for Obstetric & Gynecologic Diseases, Beijing, China.
An AI-powered diagnostic model significantly improves pelvic floor dysfunction (PFD) diagnosis by establishing new AI-Reference ranges for surface electromyography (sEMG) parameters, outperforming the older Glazer protocol.
Area of Science:
- Biomedical Engineering
- Artificial Intelligence in Medicine
- Pelvic Floor Disorders Research
Background:
- Pelvic floor surface electromyography (sEMG) is crucial for evaluating and treating pelvic floor dysfunctions (PFDs).
- The existing Glazer protocol, developed over 20 years ago with a limited sample size, faces challenges in accurately diagnosing PFDs across diverse populations.
- There is a need for updated, more reliable reference ranges and advanced diagnostic methods for PFDs.
Purpose of the Study:
- To establish a multidimensional database for monitoring pelvic floor sEMG.
- To derive more accurate AI-Reference ranges for sEMG parameters.
- To develop and validate an artificial intelligence (AI) model for the accurate diagnosis of PFDs.
Main Methods:
- A population-based, multicenter, cross-sectional study involving 1605 participants from 21 centers across China.
- Development of a multidimensional sEMG database and an AI-Diagnostician-PFD diagnostic model using AI.
- Data splitting into training (60%), testing (40%), and independent validation datasets, ensuring representation across diverse geographical regions.
Main Results:
- The AI-Reference ranges demonstrated an 11% higher area under the receiver operating characteristic curve (AUC) than the Glazer standard in the external validation dataset.
- The AI-Diagnostician-PFD model achieved AUCs of 0.81 (internal) and 0.79 (external validation), surpassing the Glazer standard's AUCs of 0.76 and 0.68.
- The AI model exhibited superior diagnostic performance for PFDs, with a 1% higher AUC compared to other classical and deep learning models.
Conclusions:
- AI-derived reference intervals significantly outperform the established Glazer standard for PFD diagnosis.
- The AI-Diagnostician-PFD model offers enhanced accuracy and will be made freely available as software.
- Implementation of this AI algorithm in clinical practice is expected to improve individual PFD diagnosis and population health outcomes.
