DOI 10.1007/s11096-024-01860-5
Developing a machine learning-based predictive model for the analgesic effectiveness of transdermal fentanyl in cancer patients: an interpretable approach.
Аннотация
BACKGROUND
Cancer-related pain is a common and distressing symptom in patients with malignant tumors, significantly affecting quality of life. Transdermal fentanyl is a convenient opioid option for patients with intestinal obstruction or difficulty swallowing; however, some patients do not experience adequate pain relief. Predicting transdermal fentanyl analgesic effectiveness is crucial to optimize pain management.
AIM: This study aimed to develop a predictive model for transdermal fentanyl effectiveness in cancer patients.
METHOD: Clinical data from adult cancer pain patients at Chongqing University Cancer Hospital were analyzed (January 2020-December 2022). Logistic regression and feature selection were applied, followed by developing nine predictive models using Logistic Regression, Random Forest (RF), and Extreme Gradient Boosting. The receiver operating characteristic (ROC) curves, the Youden index, and the Brier score were used to evaluate the performance of the model. Cross-validation and SHapley Additive exPlanations (SHAP) analysis were used for validation and feature interpretation.
RESULTS
Among 151 patients, 27.2% reported ineffectiveness of transdermal fentanyl. Logistic regression identified key factors of NRS, transdermal fentanyl dosage, BMI, and ALT. Among the nine models, RF Model 8 exhibited the best performance, achieving a ROC-AUC of 0.984 (95% CI: [0.968, 0.999]). This performance was further validated by the confusion matrix metrics and visualization results. The SHAP analysis highlighted BMI, lower doses, NRS, and ALT as predictors of transdermal fentanyl ineffectiveness.
CONCLUSION
The Random Forest model offers a valuable tool for predicting the effectiveness of transdermal fentanyl in cancer pain patients, supporting the refined assessment and management of pain.
Предметные рубрики MeSH
Подготовка и перевод публикации
Обработка статьи · этапы, ошибки и управление
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.
Тип публикации указан по данным источника. Уровень доказательности автоматически не присваивается.