<< Back to the abstract archive
Machine Learning Prediction of Surgical Site Infections Following DIEP Flap Breast Reconstruction: An NSQIP Analysis
Berk B Ozmen, MD; Diwakar Phuyal, MB BS; Graham S Schwarz, MD, MSE, FACS
Cleveland Clinic Department of Plastic Surgery
2025-01-10
Presenter: Berk B Ozmen, MD
Affidavit:
Yes
Director Name: Graham S Schwarz, MD, MSE, FACS
Author Category: Fellow Plastic Surgery
Presentation Category: Clinical
Abstract Category: Breast (Aesthetic and Recon.)
Background: Surgical site infections (SSIs) after deep inferior epigastric perforator (DIEP) flap surgery for breast reconstruction can lead to significant patient morbidity and increased healthcare costs. Accurate prediction of SSIs can enhance preoperative risk assessment and improve postoperative care. This study aims to develop and validate machine learning models to predict SSIs in patients undergoing DIEP flap surgery, thereby aiding in clinical decision-making and patient management.
Methods: We conducted a retrospective analysis using the National Surgical Quality Improvement Program (NSQIP) data from 2016 to 2022, encompassing 13,326 DIEP flap surgery cases. The dataset included 276 features such as demographic information, preoperative conditions, intraoperative variables, and postoperative outcomes. Three machine learning models; Logistic Regression, Support Vector Machine (SVM), and Random Forest were trained and evaluated.
Results: The Random Forest model outperformed the other models, achieving an accuracy of 99.66%, precision of 99.32%, recall of 100%, and an F1-score of 99.66%. The confusion matrix demonstrated robust predictive capability. Logistic Regression achieved an accuracy of 72.61%, precision of 65.88%, recall of 92.78%, and an F1-score of 77.05%. The SVM model had an accuracy of 72.29%, precision of 65.63%, recall of 92.52%, and an F1-score of 76.79%.
Conclusion: The Random Forest model demonstrated exceptional performance in predicting SSIs following DIEP flap surgery, suggesting its potential as a valuable tool for preoperative risk stratification and personalized patient care. Implementing such predictive models in clinical practice can aid surgeons in identifying high-risk patients, optimizing surgical planning, and reducing the incidence of SSIs.
