Clinical laboratory · 2026
The Application of Machine Learning in the Formation Mechanism of Eosinophilic Pleural Effusion.
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.
Yuan Xu, Huang Ming, Zhang Hui, Li Guo, Zhang Hongbo
Аннотация
BACKGROUND: The aim of this study was to establish a critical point model for the transudation of eosinophils through the pleura using machine learning techniques. This model treats the presence or absence of eosinophils in pleural effusion as a binary variable and incorporated relevant detection indicators to enhance the understanding of the formation mechanism of eosinophilic pleural effusion.
METHODS: We utilized nearly 50 indicators from 7 major detection items and compared various machine learning models, including XGBoost, Logistic Regression, LightGBM, Random Forest, AdaBoost, Decision Tree, GBDT, and GNB, to identify the optimal model based on appropriate indicators.
RESULTS: The AUC (95% CI) values for the validation sets of the XGBoost, Logistic Regression, LightGBM, Random Forest, AdaBoost, Decision Tree, GBDT, and GNB models were 0.714 (0.658 - 0.771), 0.704 (0.647 - 0.761), 0.732 (0.677 - 0.787), 0.743 (0.689 - 0.798), 0.747 (0.694 - 0.801), 0.628 (0.577 - 0.678), 0.767 (0.717 - 0.818), and 0.693 (0.637 - 0.748), respectively. Through the comparison of multiple models, the GBDT model was optimized to include the following seven indicators: APTT, TT, INR, PFTP, PTLDH, PBHGB, and DBIL. The AUC (95% CI) values for the test set were 0.902 (0.885 - 0.920), 0.761 (0.678 - 0.843), and 0.759 (0.708 - 0.811), indicating good generalization.
CONCLUSIONS: The GBDT model, identified by the code 4LE174761WZ754ba5qT1D, demonstrates that under the appropriate algorithm with the selected macroscopic indicators (APTT, TT, INR, PFTP, PTLDH, PBHGB, and DBIL), the microscopic environment is conducive to the transudation of eosinophils through the pleura, thereby optimizing the understanding of the formation mechanism of eosinophilic pleural effusion.
METHODS: We utilized nearly 50 indicators from 7 major detection items and compared various machine learning models, including XGBoost, Logistic Regression, LightGBM, Random Forest, AdaBoost, Decision Tree, GBDT, and GNB, to identify the optimal model based on appropriate indicators.
RESULTS: The AUC (95% CI) values for the validation sets of the XGBoost, Logistic Regression, LightGBM, Random Forest, AdaBoost, Decision Tree, GBDT, and GNB models were 0.714 (0.658 - 0.771), 0.704 (0.647 - 0.761), 0.732 (0.677 - 0.787), 0.743 (0.689 - 0.798), 0.747 (0.694 - 0.801), 0.628 (0.577 - 0.678), 0.767 (0.717 - 0.818), and 0.693 (0.637 - 0.748), respectively. Through the comparison of multiple models, the GBDT model was optimized to include the following seven indicators: APTT, TT, INR, PFTP, PTLDH, PBHGB, and DBIL. The AUC (95% CI) values for the test set were 0.902 (0.885 - 0.920), 0.761 (0.678 - 0.843), and 0.759 (0.708 - 0.811), indicating good generalization.
CONCLUSIONS: The GBDT model, identified by the code 4LE174761WZ754ba5qT1D, demonstrates that under the appropriate algorithm with the selected macroscopic indicators (APTT, TT, INR, PFTP, PTLDH, PBHGB, and DBIL), the microscopic environment is conducive to the transudation of eosinophils through the pleura, thereby optimizing the understanding of the formation mechanism of eosinophilic pleural effusion.