Novel approach for noninvasive pelvic floor muscle strength measurement using extracorporeal surface perineal pressure measurement and machine learning modeling
Objective Accurate measurement of pelvic floor muscle (PFM) strength is crucial for the management of pelvic floor disorders. However, the current methods are invasive, uncomfortable, and lack standardization. This study aimed to introduce a novel noninvasive approach for precise PFM strength quanti...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SAGE Publishing
2025-01-01
|
| Series: | Digital Health |
| Online Access: | https://doi.org/10.1177/20552076251316730 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Objective Accurate measurement of pelvic floor muscle (PFM) strength is crucial for the management of pelvic floor disorders. However, the current methods are invasive, uncomfortable, and lack standardization. This study aimed to introduce a novel noninvasive approach for precise PFM strength quantification by leveraging extracorporeal surface perineal pressure (ESPP) measurements and machine learning algorithms. Methods Twenty-one healthy women participated in this study. ESPP measurements were obtained using a 10 × 10 pressure array sensor during maximal voluntary PFM contractions in a seated position. Simultaneously, transabdominal ultrasound was used to measure bladder base displacement (mm) as a reference for PFM contraction strength. Seven ESPP variables were calculated based on ESPP data and intra- and inter-rater reliabilities were assessed. Machine learning algorithms predicted bladder base displacement from ESPP variables. Results The ESPP measurements demonstrated good to excellent intra-rater (ICC = 0.881) and inter-rater (ICC = 0.967) reliability. Significant correlations were observed between bladder base displacement and middle ( r = .619, P < .001) and front ( r = −.379, P =.002) vectors. The top-performing models for predicting bladder base displacement were the support vector machine [root mean square error (RMSE) = 0.139, R2 = 0.542], random forest (RMSE = 0.123, R2 = 0.367), and AdaBoost (RMSE = 0.123, R2 = 0.320) on the training set, and AdaBoost (RMSE = 0.173, R2 = 0.537), random forest (RMSE = 0.177, R2 = 0.512), and support vector machine (RMSE = 0.178, R2 = 0.508) on the test set. In predicting bladder base displacement, Bland–Altman analysis revealed these models had minimal systematic bias, with mean differences ranging from −0.007 to 0.066, and clinically acceptable limits of agreement. Conclusion This study demonstrates the potential of ESPP measurements and machine learning algorithms as a reliable and valid noninvasive approach for assessing PFM strength by quantifying the directionality of contractions, overcoming the limitations of traditional techniques. |
|---|---|
| ISSN: | 2055-2076 |