Frontiers in endocrinology · 2025
Machine learning-based predictive model for hungry bone syndrome following parathyroidectomy in secondary hyperparathyroidism.
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.
Chai Yalin, Yuan Nan, Yin Jiaming, Shen Bing, Sun Lijie, Zhang Lin, Yin Le, Wang Xuan, Luo Feng, Luo Congjuan
Аннотация
OBJECTIVE: To develop an interpretable machine learning model for predicting hungry bone syndrome (HBS) risk following parathyroidectomy in secondary hyperparathyroidism (SHPT) patients.
METHODS: This retrospective study analyzed 181 SHPT patients who underwent parathyroidectomy at the Affiliated Hospital of Qingdao University (2015 - 2025). Participants were randomly divided into a training group (70%) and a validation group (30%). From 46 candidate variables, five key predictors were selected through logistic regression and Boruta algorithm. Seven machine learning models were trained, evaluated by ROC curves, calibration curves, and decision curve analysis (DCA). Model interpretability was quantified via SHapley Additive exPlanations (SHAP).
RESULTS: The XGBoost algorithm demonstrated excellent predictive performance, with an AUC of 0.878 (95% CI: 0.779 - 0.973) and an F1 score of 0.871 for the validation cohort. The key predictors included preoperative parathyroid hormone (Pre-PTH), the percentage decay between Pre-PTH and PTH at skin closure (%PTH), alkaline phosphatase, serum calcium, and age. Additionally, we designed a web application to estimate HBS risk.
CONCLUSIONS: This interpretable machine-learning model is effective in predicting the risk of HBS in SHPT patients after parathyroidectomy, thereby providing guidance for postoperative surveillance strategies.
METHODS: This retrospective study analyzed 181 SHPT patients who underwent parathyroidectomy at the Affiliated Hospital of Qingdao University (2015 - 2025). Participants were randomly divided into a training group (70%) and a validation group (30%). From 46 candidate variables, five key predictors were selected through logistic regression and Boruta algorithm. Seven machine learning models were trained, evaluated by ROC curves, calibration curves, and decision curve analysis (DCA). Model interpretability was quantified via SHapley Additive exPlanations (SHAP).
RESULTS: The XGBoost algorithm demonstrated excellent predictive performance, with an AUC of 0.878 (95% CI: 0.779 - 0.973) and an F1 score of 0.871 for the validation cohort. The key predictors included preoperative parathyroid hormone (Pre-PTH), the percentage decay between Pre-PTH and PTH at skin closure (%PTH), alkaline phosphatase, serum calcium, and age. Additionally, we designed a web application to estimate HBS risk.
CONCLUSIONS: This interpretable machine-learning model is effective in predicting the risk of HBS in SHPT patients after parathyroidectomy, thereby providing guidance for postoperative surveillance strategies.