Related Experiment Video
Updated: May 14, 2026

Point-Of-Care Ultrasound Screening for Proximal Lower Extremity Deep Venous Thrombosis
Published on: February 10, 2023
Machine Learning Model Predicts New-Onset Lower Extremity Deep Vein Thrombosis After Pelvic Fracture Surgery and
Haoyuan Fu1,2, Qi Dong1,2, Guoqiang Li1,2
1Department of Orthopedic Surgery, Third Hospital of Hebei Medical University, Shijiazhuang, Hebei, People's Republic of China.
Machine learning models effectively predict postoperative new-onset deep vein thrombosis (PNO-DVT) after pelvic fracture surgery. XGBoost demonstrated superior performance, identifying key risk factors for early intervention.
Area of Science:
- Orthopedic Surgery
- Vascular Surgery
- Data Science in Medicine
Background:
- Postoperative new-onset deep vein thrombosis (PNO-DVT) is a significant complication following pelvic fracture surgery.
- Existing risk assessment tools for PNO-DVT lack precision.
- Machine learning (ML) offers enhanced predictive capabilities for clinical outcomes.
Purpose of the Study:
- To evaluate the efficacy of various ML models in predicting PNO-DVT after pelvic fracture surgery.
- To identify independent risk factors for PNO-DVT using ML approaches.
- To compare the predictive performance of different ML algorithms.
Main Methods:
- Analysis of clinical data from 745 patients undergoing pelvic fracture surgery (2016-2019).
- Identification of 12 independent risk factors using logistic and LASSO regression.
- Development and evaluation of six ML models: logistic regression, SVM, random forest, XGBoost, LightGBM, and AdaBoost.
Main Results:
- XGBoost achieved the highest Area Under the Curve (AUC) of 0.8633, indicating superior predictive performance.
- Key independent risk factors identified include age, BMI, intraoperative blood loss, and HDL-C levels.
- Model accuracy varied across algorithms, ranging from 0.6502 to 0.9283.
Conclusions:
- Machine learning, particularly XGBoost, effectively predicts PNO-DVT in patients with pelvic fractures.
- Identified predictors like age, intraoperative blood loss, and BMI facilitate accurate risk stratification.
- These findings support the implementation of early preventive strategies for PNO-DVT.
Related Concept Videos
Venous Thrombosis III: Interprofessional Care
Venous Thrombosis II: Clinical Manifestations and Diagnostic Studies
Pulmonary Embolism III: Nursing Management
Venous Thrombosis IV: Nursing Management
Peripheral Artery Disease V: Postoperative Nursing Management
Venous Thrombosis I: Introduction