European journal of surgical oncology : the journal of the European Society of Surgical Oncology and the British Association of Surgical Oncology · 2026
Development and validation of an ensemble machine learning model to predict survival in locally advanced rectal cancer: A multicenter, retrospective study.
СтатьяМногоцентровое исследованиеНаблюдательное исследованиеВалидационное 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.
Pan Zhen, Zheng Shaoqing, Zhuang Zhicheng, Wang Ye, Lu XingRong, Peng Tengyi, Zhang Qiqi, Ye Wen, Guan Guoxian, Li Shoufeng, Chen Bin
Аннотация
BACKGROUND: Patients identified with locally advanced rectal cancer (LARC) demonstrate a varied prognosis after undergoing neoadjuvant chemoradiotherapy (nCRT), highlighting the essential need for precise predictions of outcomes.
PURPOSE: The aim of this study is to create and assess a machine learning model that is interpretable and tailored to forecasting results in individuals diagnosed with LARC.
METHODS: A multicenter retrospective cohort study was carried out, incorporating 1119 instances of LARC that received radical surgery following nCRT between the years 2012 and 2022. We utilized ten feature selection machine learning algorithms to identify the optimal predictive factors. Subsequently, we developed models using the selected subset of ten features combined with ten machine learning algorithms. The models' effectiveness was assessed using two distinct cohorts and analyzed through multiple techniques, such as the time-dependent calibration curves, concordance index (C-index), decision curve analysis and time-dependent receiver operating characteristic curves.
RESULTS: Following the selection of predictors, a total of ten feature subsets were created. These subsets were then paired with ten machine learning algorithms in various combinations, leading to the formation of 100 predictive models. Of all the models analyzed, the integration of Random Survival Forest with gradient boosting showed the highest level of predictive accuracy. In the training group, the C-index for GRM was recorded at 0.917 (95% CI 0.890-0.944), while in validation cohort 1, it was 0.897 (95% CI 0.850-0.924), and in validation cohort 2, it registered at 0.837 (95% CI 0.780-0.894). Moreover, a web-based tool that is accessible to the public was developed for the GRM.
CONCLUSION: GRM possesses the capability to effectively determine the prognosis for patients with LARC undergoing nCRT. This can aid healthcare providers in assessing the severity of the condition, improve patient oversight, and assist in the development of supplementary treatment strategies.
PURPOSE: The aim of this study is to create and assess a machine learning model that is interpretable and tailored to forecasting results in individuals diagnosed with LARC.
METHODS: A multicenter retrospective cohort study was carried out, incorporating 1119 instances of LARC that received radical surgery following nCRT between the years 2012 and 2022. We utilized ten feature selection machine learning algorithms to identify the optimal predictive factors. Subsequently, we developed models using the selected subset of ten features combined with ten machine learning algorithms. The models' effectiveness was assessed using two distinct cohorts and analyzed through multiple techniques, such as the time-dependent calibration curves, concordance index (C-index), decision curve analysis and time-dependent receiver operating characteristic curves.
RESULTS: Following the selection of predictors, a total of ten feature subsets were created. These subsets were then paired with ten machine learning algorithms in various combinations, leading to the formation of 100 predictive models. Of all the models analyzed, the integration of Random Survival Forest with gradient boosting showed the highest level of predictive accuracy. In the training group, the C-index for GRM was recorded at 0.917 (95% CI 0.890-0.944), while in validation cohort 1, it was 0.897 (95% CI 0.850-0.924), and in validation cohort 2, it registered at 0.837 (95% CI 0.780-0.894). Moreover, a web-based tool that is accessible to the public was developed for the GRM.
CONCLUSION: GRM possesses the capability to effectively determine the prognosis for patients with LARC undergoing nCRT. This can aid healthcare providers in assessing the severity of the condition, improve patient oversight, and assist in the development of supplementary treatment strategies.