Related Experiment Video
Updated: Jul 4, 2026
In Vitro Microfluidic Disease Model to Study Whole Blood-Endothelial Interactions and Blood Clot Dynamics in Real-Time
Published on: May 24, 2020
Development and validation of a novel orthopedic blood use prediction model incorporating lab tests and
Aqin Chen1, Bing Cui1, Chengcheng Xu2
1Department of Blood Transfusion, Affiliated Hospital of Jiangsu University, China.
Objective:
To develop and validate a prediction model integrating laboratory parameters and thromboelastography (TEG) for forecasting blood transfusion needs in orthopedic surgery.
Methods:
This retrospective study enrolled 250 patients undergoing joint replacement, spinal fusion, or fracture fixation. Participants were randomized into training (n = 175) and validation (n = 75) sets. Preoperative demographics, laboratory indices, and TEG parameters were collected. Potential predictors were identified through univariate analysis and Least Absolute Shrinkage and Selection Operator (LASSO) regression. Significant variables were incorporated into multivariate logistic regression to identify independent factors. Multiple machine learning models-random forest (RF), k-nearest neighbors (KNN), and gradient boosting machine (GBM)-were constructed and evaluated using the area under the receiver operating characteristic curve (AUC).
Results:
No significant differences were observed in baseline characteristics between the training and validation sets (P > 0.05). Univariate analysis revealed statistically significant differences between the non-transfusion and transfusion groups in the training set regarding preoperative Hb, PLT, TEG-R, TEG-K, TEG-MA, TEG-LY30, prothrombin time, and D-dimer levels (all P < 0.05). Multivariate logistic regression analysis demonstrated that preoperative Hb, preoperative PLT, and TEG-MA were independent protective factors against postoperative transfusion (all P < 0.05), whereas TEG-R, TEG-K, TEG-LY30, and preoperative prothrombin time were independent risk factors (all P < 0.05). The random forest model achieved the highest AUC (0.931), significantly outperforming the KNN (0.894) and GBM (0.813) models, thus being selected as the optimal predictive model.
Conclusion:
The random forest model based on laboratory parameters and TEG parameters effectively predicts postoperative transfusion requirements in orthopedic surgery patients, providing an objective basis for preoperative blood risk assessment and individualized transfusion strategies.
