Predicting Surgical Outcomes in Breast Reconstruction With Machine Learning: A Systematic Review.

Area of Science:

• Plastic Surgery
• Artificial Intelligence
• Machine Learning

Background:

• Artificial intelligence (AI) and machine learning (ML) are increasingly used in plastic surgery to predict patient outcomes and guide decision-making.
• This review focuses on the performance of ML models specifically within breast reconstruction.

Purpose of the Study:

• To systematically review and evaluate the performance of machine learning (ML) prediction models in breast reconstruction.
• To compare the effectiveness of different ML models and outcome measures in predicting surgical results.

Main Methods:

• A systematic review of PubMed, Scopus, and EMBASE was performed.
• Included studies used ML to predict outcomes in breast reconstruction, reporting model types and performance metrics (e.g., area under the receiver operating characteristic curve).
• Statistical analyses included descriptive statistics, multivariate linear regression, and meta-regression.

Main Results:

• Fourteen studies involving 19 ML models and 11,013 patients were analyzed.
• The median area under the receiver operating characteristic curve across all models was 0.71.
• Models predicting BREAST-Q outcomes and those using class imbalance mitigation showed significantly higher discrimination.

Conclusions:

• Machine learning models are effective for predicting various outcomes in breast reconstruction, including surgical complications and patient satisfaction.
• Models predicting BREAST-Q and employing class imbalance methods demonstrated superior discrimination.
• Standardized reporting is crucial for future ML applications in plastic surgery to ensure reproducibility and facilitate comparisons.