Key result
The XGBoost machine learning model demonstrated the best predictive performance for venous thromboembolism among hospitalized cancer patients, achieving an AUROC of 0.818.
Why the study?
Hospitalized cancer patients are at high risk for venous thromboembolism, making accurate risk identification on admission vital, and machine learning methods may aid model development.
Does an XGBoost machine learning model accurately predict the risk of venous thromboembolism in hospitalized cancer patients?
Case-Control (n=1,100)
No
Does an XGBoost machine learning model accurately predict the risk of venous thromboembolism in hospitalized cancer patients?
An XGBoost machine learning model utilizing clinical and laboratory features on admission can effectively predict the risk of venous thromboembolism in hospitalized cancer patients.
Answers are grounded in this paper and cite their sources.
ML models may aid VTE risk stratification in cancer inpatients; leaves open prospective validation and outcome impact versus standard tools.
Hospitalized cancer patients are high-risk venous thromboembolism (VTE) population. It is vital to identify the VTE risk of inpatients on admission accurately. Machine learning (ML) has advantages in data processing and model development. The article aims to develop predictive models using four ML methods to evaluate VTE risk for hospitalized cancer patients and select a model with the best performance. A retrospective cohort study was conducted for cancer patients hospitalized at Hunan Cancer Hospital between December 1, 2017 and November 30, 2020. A total of 1,100 inpatients were included in the study. There were 340 patients in the VTE group and 760 patients in the non-VTE group. The XGBoost model had the best performance for the prediction of VTE risk compared to the other three machine learning models. The AUROC value of the XGBoost model was 0.818(95%CI: 0.762, 0.870). The top five important features in the XGBoost model were D-dimer, diabetes, hypertension, pleural metastasis and hematological malignancies. Machine learning models contribute to evaluating the VTE risk in hospitalized cancer patients. They may be the practical risk stratification tools to provide a clinical reference for clinicians and nurses, especially the XGBoost model.
No takes yet. Share an insight, caveat, or question.
Meng et al. (2022) conducted a case-control in Venous thromboembolism in hospitalized cancer patients (n=1,100). XGBoost machine learning model vs. Logistic Regression, Support Vector Machine, and Random Forest models was evaluated on Area under the receiver operating characteristic curve (AUROC) for VTE prediction (AUROC 0.818, 95% CI 0.762-0.870). The XGBoost machine learning model demonstrated the best predictive performance for venous thromboembolism among hospitalized cancer patients, achieving an AUROC of 0.818.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: