Related Experiment Video
Updated: Jun 28, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.8K
Prediction models for postoperative recurrence of non-lactating mastitis based on machine learning
Jiaye Sun1, Shijun Shao1, Hua Wan2
1Department of Mammary, Shuguang Hospital, Shanghai University of Traditional Chinese Medicine, 200021, Shanghai, China.
BMC Medical Informatics and Decision Making
|April 22, 2024
Summary
Machine learning models predict non-lactating mastitis recurrence risk using Random Forest and XGBoost. XGBoost demonstrated superior performance, identifying key predictors like intraoperative discharge for better clinical guidance.
Area of Science:
- Medical Informatics
- Machine Learning in Healthcare
- Surgical Oncology
Background:
- Postoperative non-lactating mastitis (NLM) recurrence poses a clinical challenge.
- Predictive models are needed to identify patients at high risk for recurrence.
- Machine learning offers potential for developing accurate predictive tools.
Purpose of the Study:
- To develop and compare Random Forest (RF) and XGBoost machine learning models for predicting NLM recurrence.
- To identify key clinical features influencing NLM recurrence risk.
- To provide clinical guidance for treatment planning in NLM patients.
Main Methods:
- Retrospective analysis of 258 inpatient NLM cases from July 2019 to December 2021.
- Ten selected features (e.g., BMI, WBC, intraoperative discharge) used to train RF and XGBoost models.
- Model performance evaluated using Accuracy, Precision, Recall, F1-score, and AUC; SHAP values used for interpretation.
Main Results:
- A recurrence rate of 18.6% was observed in the NLM patient cohort.
- XGBoost model outperformed the RF model in predicting NLM recurrence.
- Intraoperative discharge was the most significant predictor in the XGBoost model; BMI in the RF model.
Conclusions:
- Machine learning models, particularly XGBoost, demonstrate significant predictive ability for NLM recurrence.
- SHAP value analysis validates model predictions and enhances clinical interpretability.
- These models offer valuable tools for risk stratification and personalized treatment strategies in NLM management.
