Prediction of Lymph Node Metastasis in Stage I-III Colorectal Cancer Patients Using Machine Learning Algorithms

Background Lymph node status significantly impacts patient prognosis and therapeutic decision-making. This study aimed to develop a machine learning (ML)-based predictive model for lymph node metastasis (LNM) in stage I-III colorectal cancer (CRC) patients. Methods Data were retrospectively collected from CRC patients who underwent surgical resection (curative or radical surgery) at Enze Hospital, Taizhou, Zhejiang Province, China (January 1, 2015–December 31, 2019). The least absolute shrinkage and selection operator (LASSO) regression was used for feature selection. Six ML algorithms were employed to construct predictive models. Model performance was evaluated based on discrimination, calibration, and clinical utility. The Shapley Additive Explanations (SHAP) method was applied for feature interpretation. Results Among 398 CRC patients, 158 (39.7%) had LNM. The random forest (RF) model demonstrated optimal performance, achieving an area under the curve (AUC) of 0.868 in the training set and 0.794 in the validation set. Key predictors included perineural invasion (PNI), lymphovascular invasion (LVI), tumor deposits, tumor gross type, and tumor size. Conclusions By using the tumor markers and pathological features readily available in the case data, the RF machine learning model can accurately predict the risk of lymph node metastasis in patients with stage I-III colorectal cancer, providing a potential tool for preoperative staging and personalized postoperative treatment.