Australian critical care : official journal of the Confederation of Australian Critical Care Nurses · 2026
Exploratory prediction model for deep vein thrombosis in intensive care patients after anticoagulation therapy: An observational study.
СтатьяНаблюдательное исследованиеНазвание 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.
Lv Mengjiao, Wei Yanlong, Jiang Junjie, Wang Saisai, Zhang Changyue, Qian Chunting, Zhang Chengwei, Zheng Ziqiu, Wei Rui, Xing Xiaohong, Guo Lirong
Аннотация
BACKGROUND: Despite low-molecular-weight heparin (LMWH) prophylaxis, the incidence of deep vein thrombosis (DVT) remains high in intensive care unit (ICU) patients, creating a need for personalised risk assessment to enable precision prevention.
OBJECTIVES: The aim of this study was to develop and validate a prediction model for lower extremity DVT in ICU patients already receiving LMWH, thereby improving risk stratification to guide targeted screening and optimise preventive strategies.
METHODS: In this retrospective case-control study, 601 ICU patients from a Chinese tertiary centre (Jan 2023-May 2025) were analysed. Using a time-series split, patients were divided into training (n = 459) and validation (n = 142) sets. The model was developed by screening 80 predictors through a univariate analysis (P < 0.05), followed by feature selection using LASSO regression with 10‑fold cross-validation to optimize the penalty parameter λ for optimal variable reduction, and finalised with multivariate logistic regression. The optimal threshold, determined by maximising the F1-score on the training set, was applied to the validation set to evaluate class-specific performance. Discrimination, calibration, and clinical utility were assessed in both sets using receiver operating characteristic curves, the Hosmer-Lemeshow test, and the decision curve analysis, respectively. Model stability was further assessed by comparing its receiver operating characteristic curve with those of other machine learning models.
RESULTS: The final model incorporated seven key predictors-surgery, tracheostomy with mechanical ventilation, central venous catheterisation, activated partial thromboplastin time, high-density lipoprotein cholesterol, glucocorticoid use, and age. At the optimal threshold of 0.48, it demonstrated high sensitivity (82.2%) for DVT screening and high precision (90.8%) for ruling out DVT, providing effective triage capability. The model showed strong discriminative power, with an area under the curve >0.75, which was consistent across multiple machine learning models, including multivariate logistic regression. The Hosmer-Lemeshow test indicated good fit in both training and validation sets (P > 0.05). Decision curve analysis confirmed the model's superior clinical utility, showing a higher net benefit than "treat-all" or "treat-none" strategies across risk threshold probabilities of 0-0.6 in the training set and 0-0.8 in the validation set.
CONCLUSIONS: This exploratory prediction model, based on seven routinely available clinical variables, provides preliminary risk stratification for DVT in ICU patients receiving LMWH prophylaxis. However, this model requires external validation before clinical application and should be considered hypothesis generating at this stage.
OBJECTIVES: The aim of this study was to develop and validate a prediction model for lower extremity DVT in ICU patients already receiving LMWH, thereby improving risk stratification to guide targeted screening and optimise preventive strategies.
METHODS: In this retrospective case-control study, 601 ICU patients from a Chinese tertiary centre (Jan 2023-May 2025) were analysed. Using a time-series split, patients were divided into training (n = 459) and validation (n = 142) sets. The model was developed by screening 80 predictors through a univariate analysis (P < 0.05), followed by feature selection using LASSO regression with 10‑fold cross-validation to optimize the penalty parameter λ for optimal variable reduction, and finalised with multivariate logistic regression. The optimal threshold, determined by maximising the F1-score on the training set, was applied to the validation set to evaluate class-specific performance. Discrimination, calibration, and clinical utility were assessed in both sets using receiver operating characteristic curves, the Hosmer-Lemeshow test, and the decision curve analysis, respectively. Model stability was further assessed by comparing its receiver operating characteristic curve with those of other machine learning models.
RESULTS: The final model incorporated seven key predictors-surgery, tracheostomy with mechanical ventilation, central venous catheterisation, activated partial thromboplastin time, high-density lipoprotein cholesterol, glucocorticoid use, and age. At the optimal threshold of 0.48, it demonstrated high sensitivity (82.2%) for DVT screening and high precision (90.8%) for ruling out DVT, providing effective triage capability. The model showed strong discriminative power, with an area under the curve >0.75, which was consistent across multiple machine learning models, including multivariate logistic regression. The Hosmer-Lemeshow test indicated good fit in both training and validation sets (P > 0.05). Decision curve analysis confirmed the model's superior clinical utility, showing a higher net benefit than "treat-all" or "treat-none" strategies across risk threshold probabilities of 0-0.6 in the training set and 0-0.8 in the validation set.
CONCLUSIONS: This exploratory prediction model, based on seven routinely available clinical variables, provides preliminary risk stratification for DVT in ICU patients receiving LMWH prophylaxis. However, this model requires external validation before clinical application and should be considered hypothesis generating at this stage.