Machine learning models predict lymph node metastasis in patients with stage T1-T2 esophageal squamous cell carcinoma

Area of Science:

• Oncology
• Medical Informatics
• Machine Learning

Background:

• Accurate prediction of lymph node metastasis (LNM) is crucial for stage T1-T2 esophageal squamous cell carcinoma (ESCC).
• Existing methods for LNM prediction in early-stage ESCC present challenges.

Purpose of the Study:

• To evaluate the performance of machine learning (ML) models in predicting LNM for patients with stage T1-T2 ESCC.
• To identify the optimal ML model for LNM prediction in this patient cohort.

Main Methods:

• A retrospective study included 1097 patients with stage T1-T2 ESCC.
• Thirty-six ML models were developed using various algorithms and feature selection techniques.
• Model performance was assessed using the area under the receiver operating characteristic curve (AUC) on training and external test sets.

Main Results:

• ML models demonstrated good predictive performance, with a median bootstrapped AUC of 0.659.
• The optimal model, a naive Bayes algorithm with determination coefficient feature selection, achieved an AUC of 0.715 in the training set.
• This optimal ML model showed superior performance in the external test set with an AUC of 0.752, outperforming T stage prediction (AUC 0.624).

Conclusions:

• Machine learning models offer significant value in predicting LNM for stage T1-T2 ESCC.
• The naive Bayes algorithm, combined with determination coefficient feature selection, emerged as the best-performing model for LNM prediction.