International journal of surgery (London, England) · 2026
Development and internal-external validation of a machine learning model to predict the risk for postoperative adverse events in patients with peptic ulcer perforation: a secondary cohort study of the perforated peptic ulcer analyzing project study.
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.
Ito Kei, Endo Akira, Hoshi Hiromasa, Ito Koji, Akutsu Tomohiro, Odera Hikaru, Shiraki Hideto, Yokoyama Takeshi, Narita Yasukazu, Masuda Taro, Suekane Akira, Yamagishi Shigeru, Morishita Koji
Аннотация
BACKGROUND: Morbidity and mortality rates after surgery in patients with perforated peptic ulcer (PPU) remain high. Although several scoring systems have been proposed, only a few machine learning models have been designed to predict postoperative adverse events. This study aimed to develop and deploy a predictive model for the risk of postoperative adverse events in patients with PPU.
METHODS: We analyzed data from the Perforated Peptic Ulcer Analyzing Project, which is a retrospective survey of adult patients with PPU in seven participating institutions in Japan between January 2011 and December 2022. The postoperative adverse events were defined as complications with a Clavien-Dindo classification of Grade III or higher. The variables were selected based on permutation feature importance derived from a random forest algorithm using bootstrap sampling. Internal-external cross-validation was conducted to develop and evaluate the model's performance. The final model was deployed as an interactive web application.
RESULTS: Of the 702 potentially eligible patients, 425 were treated surgically. Of these, 78 patients (18.3%) experienced postoperative adverse events. Ten variables were selected and used to develop a random forest model that displayed favorable discriminatory ability, with a pooled area under the receiver operating characteristic curve (AUROC) of 0.83. Compared to the other systems, the random forest model outperformed the PULP (AUROC, 0.79) and Boey scores (AUROC, 0.67).
CONCLUSION: In this study, a machine learning model was developed and deployed to predict postoperative adverse events in patients with PPU. Further external validation is required for clinical use.
METHODS: We analyzed data from the Perforated Peptic Ulcer Analyzing Project, which is a retrospective survey of adult patients with PPU in seven participating institutions in Japan between January 2011 and December 2022. The postoperative adverse events were defined as complications with a Clavien-Dindo classification of Grade III or higher. The variables were selected based on permutation feature importance derived from a random forest algorithm using bootstrap sampling. Internal-external cross-validation was conducted to develop and evaluate the model's performance. The final model was deployed as an interactive web application.
RESULTS: Of the 702 potentially eligible patients, 425 were treated surgically. Of these, 78 patients (18.3%) experienced postoperative adverse events. Ten variables were selected and used to develop a random forest model that displayed favorable discriminatory ability, with a pooled area under the receiver operating characteristic curve (AUROC) of 0.83. Compared to the other systems, the random forest model outperformed the PULP (AUROC, 0.79) and Boey scores (AUROC, 0.67).
CONCLUSION: In this study, a machine learning model was developed and deployed to predict postoperative adverse events in patients with PPU. Further external validation is required for clinical use.