Analysis and application of a machine learning-based assisted diagnosis method for myofascial pelvic pain syndrome

Objectives: The research aims to leverage machine learning techniques to better understand the diagnosis of myofascial pelvic pain syndrome (MPPS) and to develop useful tools for clinical practice. Methods: , we developed prediction models with 10 machine learning algorithms: logistic regression, support vector machine (SVM), decision tree (DT), random forest (RF), eXtreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), adaptive boosting (AdaBoost), categorical boosting (CatBoost), k-nearest neighbors (KNN), and backpropagation (BP). Five-fold cross-validation was used to prevent overfitting. The models' performance was evaluated using accuracy, precision, recall, F1 score, and the area under the receiver operating characteristic curve (AUC-ROC) to assess each algorithm's diagnostic value for MPPS. Results: The top four models in terms of AUC, ranked from highest to lowest, were RF, CatBoost, XGBoost, and LightGBM. The top four models in terms of accuracy, ranked from highest to lowest, were CatBoost, RF, XGBoost, and LightGBM. Moreover, the top four models in terms of area under the decision curve (AUDC), ranked from highest to lowest, were CatBoost, LightGBM, XGBoost, and RF. Furthermore, we created a web-based graphical user interface (GUI) for MPPS prediction. It can be packaged for cross-platform use, thereby streamlining diagnosis and improving accessibility for healthcare providers. Conclusion: In conclusion, this study compared 10 machine learning algorithms for diagnosing myofascial pelvic pain syndrome. The CatBoost model showed superior performance in terms of accuracy and clinical utility. In addition, a cross-platform web-based GUI was developed, streamlining diagnosis for healthcare providers and potentially improving patient outcomes.