International journal of medical sciences · 2026
Machine Learning for Lymph Node Metastasis Prediction in Early Gastric Cancer: A Comparative Analysis.
Сравнительное исследованиеСтатьяВалидационное исследованиеСравнительное исследованиеНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Chen Yufan, Bai Kunhao, Yang Minghui, Ma Chao, Gao Xiaohang, Xu Guoliang, Chen Yingbo, Zhang Rong
Аннотация
OBJECTIVE: Lymph node metastasis (LNM) plays a crucial role in informing treatment decisions and prognosis for early gastric cancer (EGC). This study aimed to offer a practical approach to predict LNM in EGC by using machine learning algorithms.
METHODS: This study collected data from 1085 patients with EGC who underwent radical gastrectomy with D1+ or D2 lymph node resection. Seven machine-learning algorithms were compared, and hyperparameters were fine-tuned to identify the model with the best accuracy, Brier class and Area Under the Curve (AUC). The efficacy of the selected model was evaluated.
RESULTS: Following comparison, the Random Forest (RF), Extreme Gradient Boosting (Boost), and Neural Network (NNT) models exhibited exemplary performance on the training dataset, with AUC values of 0.796, 0.788, and 0.779, respectively, on the validation set. We conducted parallel analyses within the T1a and T1b subgroups, where Logistics Models (LM) and RF yielded AUCs of 0.710 and 0.636 in the T1a validation set, and LM, RF, and Boost achieved AUCs of 0.666, 0.658, and 0.558, respectively in the T1b validation set. Variable importance analysis utilizing SHAP revealed distinct values for lymph node metastasis (LNM) in EGC patients, as well as in those stratified into T1a and T1b groups.
CONCLUSION: The machine learning model holds the potential to guide more effective treatment strategies for early gastric cancer (EGC), specifically in addressing lymph node metastasis (LNM). The identified risk factors contribute valuable insights for personalized decision-making in the management of EGC patients.
METHODS: This study collected data from 1085 patients with EGC who underwent radical gastrectomy with D1+ or D2 lymph node resection. Seven machine-learning algorithms were compared, and hyperparameters were fine-tuned to identify the model with the best accuracy, Brier class and Area Under the Curve (AUC). The efficacy of the selected model was evaluated.
RESULTS: Following comparison, the Random Forest (RF), Extreme Gradient Boosting (Boost), and Neural Network (NNT) models exhibited exemplary performance on the training dataset, with AUC values of 0.796, 0.788, and 0.779, respectively, on the validation set. We conducted parallel analyses within the T1a and T1b subgroups, where Logistics Models (LM) and RF yielded AUCs of 0.710 and 0.636 in the T1a validation set, and LM, RF, and Boost achieved AUCs of 0.666, 0.658, and 0.558, respectively in the T1b validation set. Variable importance analysis utilizing SHAP revealed distinct values for lymph node metastasis (LNM) in EGC patients, as well as in those stratified into T1a and T1b groups.
CONCLUSION: The machine learning model holds the potential to guide more effective treatment strategies for early gastric cancer (EGC), specifically in addressing lymph node metastasis (LNM). The identified risk factors contribute valuable insights for personalized decision-making in the management of EGC patients.
Предметные рубрики MeSH
Machine LearningPredictive Learning ModelsLymphatic MetastasisStomach NeoplasmsRandom ForestBoosting Machine Learning AlgorithmsNeural Networks, ComputerArea Under CurveNeoplasm StagingPrognosisRisk FactorsGastrectomyLymph Node ExcisionElectronic Health RecordsClinical Decision-MakingHumansMaleFemaleAdultMiddle AgedAgedRisk Assessment