Machine learning-enabled prediction of bone metastasis in esophageal cancer.

Purpose: Bone metastasis (BM) is a common manifestation of distant spread in patients with esophageal cancer. This study aimed to develop a machine learning algorithm to predict the risk of bone metastasis in esophageal cancer patients, thereby supporting clinical decision-making support. Methods: Clinical and pathological data of esophageal cancer patients were obtained from the SEER database of the U.S. National Institutes of Health from 2010 to 2020. Six machine learning models were constructed: Support Vector Machine, Logistic Regression, Extreme Gradient Boosting, Neural Network, Random Forest, and k-Nearest Neighbors. Models performance was evaluated using accuracy, precision, recall, F1-score, and the area under the receiver operating characteristic curve. The optimal model was further used to interpret the associations between clinicopathological features and bone metastasis. Results: A total of 9,744 patients were included, with 532 (5.47%) had bone metastasis and 9,212 (94.53%) without. Multivariate logistic regression analysis identified age, T stage, N stage, and histological type as independent risk factors for bone metastasis. The XGBoost model demonstrated the best performance, achieving an accuracy of 0.80, a recall of 0.99, a precision of 0.72, an F1-score of 0.8300, and AUC of 0.92. Conclusion: The XGBoost model showed excellent predictive performance for bone metastasis in esophageal cancer patients, providing valuable insights for guiding clinical treatment decisions.