Predicting periprosthetic joint Infection: Evaluating supervised machine learning models for clinical application.

Area of Science:

• Orthopedic surgery
• Medical informatics
• Machine learning in healthcare

Background:

• Periprosthetic joint infection (PJI) is a severe complication following joint arthroplasty.
• PJI leads to significant patient morbidity, increased costs, and reduced quality of life.
• Accurate prediction of PJI is crucial for effective patient management.

Purpose of the Study:

• To evaluate the predictive performance of various supervised machine learning models for PJI.
• To identify the most effective models for clinical application in predicting PJI using patient data.

Main Methods:

• Trained and tested eight supervised machine learning models (Logistic Regression, Random Forest, XGBoost, ANN, KNN, AdaBoost, GNB, SGD).
• Utilized a dataset of 27,854 patients who underwent joint arthroplasty.
• Evaluated models using accuracy, precision, recall, specificity, F1 score, and AUC.

Main Results:

• Random Forest and XGBoost demonstrated superior performance with high accuracy and balanced metrics.
• k-Nearest Neighbors (KNN) also showed strong results, particularly in minimizing misclassifications.
• Gaussian Naive Bayes (GNB) and Stochastic Gradient Descent (SGD) exhibited weaker performance with higher error rates.

Conclusions:

• Random Forest, XGBoost, and KNN are promising for clinical implementation in PJI prediction.
• These models can potentially support earlier diagnosis and improve patient outcomes in orthopedic surgery.
• Machine learning models offer a valuable tool for enhancing clinical decision-making and reducing infection-related morbidity.