BMC gastroenterology · 2025
Diagnosis model for gastric submucosal tumor based on multiple decision trees comprising endoscopic and endoscopic ultrasonography features.
СтатьяНазвание и аннотация: Перевод в очереди
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.
Su Shen, Wu Shun-Yao, Tang Yun-He, Ren Lin-Lin, Yin Fan, Xu Yu-Shuang, Li Xiao-Yu, Liu Hua, Zhang Shao-Hua, Zhang Xing-Lin, Tian Zi-Bin, Mao Tao
Аннотация
BACKGROUND: The clinical course and manifestations of subepithelial lesions (SEL) patients tend to vary from different pathological types, therefore the accurate diagnosis of SEL would undoubtedly be beneficial to the treatment and prognosis of SEL patients. Based on the decision tree method, we developed a novel classification model for SELs by combining endoscopy with endoscopic ultrasound (EUS).
METHODS: We retrospectively collected data from 469 patients hospitalized in the Affiliated Hospital of Qingdao University between January 2017 to November 2021 for endoscopic resection. Chi-square test (P < 0.05), independence test (P < 0.001), and Pearson correlation analysis (|r|<0.8) were performed to identify significant variables among endoscopic and EUS features, which were subsequently incorporated into decision tree analysis. Finally, a hierarchical diagnostic model based on multiple decision trees was constructed. The predictive performance of the model was obtained through a five-fold cross-validation, and each decision tree model was evaluated by the area under the curve (AUC) and F1.
RESULTS: A total of 13 variables were included in the construction of the model. The overall accuracy of this hierarchical model was 75.12%. The AUC values for each pathology type, namely gastrointestinal stromal tumor (GIST) and schwannoma, leiomyoma, inflammatory fibroid polyp, heterotopic pancreas, and lipoma, were 0.882, 0.866, 0.964, 0.863, and 0.953, respectively. And F1 of them were 0.777, 0.697, 0.658, 0.904, and 0.698, respectively.
CONCLUSIONS: This decision tree-based hierarchical model can potentially assist in the preoperative diagnosis of SEL and guide clinical decision-making for the individualized treatment of SEL patients.
METHODS: We retrospectively collected data from 469 patients hospitalized in the Affiliated Hospital of Qingdao University between January 2017 to November 2021 for endoscopic resection. Chi-square test (P < 0.05), independence test (P < 0.001), and Pearson correlation analysis (|r|<0.8) were performed to identify significant variables among endoscopic and EUS features, which were subsequently incorporated into decision tree analysis. Finally, a hierarchical diagnostic model based on multiple decision trees was constructed. The predictive performance of the model was obtained through a five-fold cross-validation, and each decision tree model was evaluated by the area under the curve (AUC) and F1.
RESULTS: A total of 13 variables were included in the construction of the model. The overall accuracy of this hierarchical model was 75.12%. The AUC values for each pathology type, namely gastrointestinal stromal tumor (GIST) and schwannoma, leiomyoma, inflammatory fibroid polyp, heterotopic pancreas, and lipoma, were 0.882, 0.866, 0.964, 0.863, and 0.953, respectively. And F1 of them were 0.777, 0.697, 0.658, 0.904, and 0.698, respectively.
CONCLUSIONS: This decision tree-based hierarchical model can potentially assist in the preoperative diagnosis of SEL and guide clinical decision-making for the individualized treatment of SEL patients.