Predicting Postoperative Recurrence Using a Support Vector Machine for Patients With Esophageal Squamous Cell...

Area of Science:

• Oncology
• Machine Learning in Medicine
• Surgical Outcomes Research

Background:

• Esophageal squamous cell carcinoma (ESCC) survival prediction models are common, but postoperative recurrence prediction is less developed.
• Accurate prediction of recurrence is crucial for effective management of ESCC patients post-surgery.

Purpose of the Study:

• To develop and validate a support vector machine (SVM)-based model for predicting postoperative recurrence risk in ESCC.
• To identify key factors associated with recurrence in ESCC patients following surgical intervention.

Main Methods:

• Retrospective analysis of clinical data from 311 ESCC patients undergoing surgery.
• Development and validation of SVM algorithms to stratify patients into high- or low-recurrence-risk groups.
• Performance evaluation using sensitivity, specificity, Youden index, and calibration curves across test and validation cohorts.

Main Results:

• The SVM7 model, incorporating TNM stage, adjuvant therapy, differentiation, tumor size, and complications, showed superior recurrence prediction sensitivity.
• A composite model (SVM6+8) achieved high prediction sensitivities (up to 94%) and specificities across cohorts.
• SVM-based risk stratification correlated with significantly longer disease-free survival, highlighting its clinical utility.

Conclusions:

• The developed SVM-based model provides accurate prediction of postoperative recurrence in ESCC patients.
• The model demonstrates high sensitivity, specificity, and discriminative power for clinical risk stratification.
• This tool aids clinicians in identifying patients at high risk for recurrence, facilitating personalized treatment strategies.