Artificial neural networks for the prediction of transfusion rates in primary total hip arthroplasty
Wayne Brian Cohen-Levy1, Christian Klemt1, Venkatsaiakhil Tirumala1
1Bioengineering Department, Department of Orthopaedic Surgery, Massachusetts General Hospital, Harvard Medical School, 55 Fruit Street, Boston, MA, 02114, USA.
Background:
Despite advancements in total hip arthroplasty (THA) and the increased utilization of tranexamic acid, acute blood loss anemia necessitating allogeneic blood transfusion persists as a post-operative complication. The prevalence of allogeneic blood transfusion in primary THA has been reported to be as high as 9%. Therefore, this study aimed to develop and validate novel machine learning models for the prediction of transfusion rates following primary total hip arthroplasty.
Methods:
A total of 7265 consecutive patients who underwent primary total hip arthroplasty were evaluated using a single tertiary referral institution database. Patient charts were manually reviewed to identify patient demographics and surgical variables that may be associated with transfusion rates. Four state-of-the-art machine learning algorithms were developed to predict transfusion rates following primary THA, and these models were assessed by discrimination, calibration, and decision curve analysis.
Results:
The factors most significantly associated with transfusion rates include tranexamic acid usage, bleeding disorders, and pre-operative hematocrit (< 33%). The four machine learning models all achieved excellent performance across discrimination (AUC > 0.78), calibration, and decision curve analysis.
Conclusion:
This study developed machine learning models for the prediction of patient-specific transfusion rates following primary total hip arthroplasty. The results represent a novel application of machine learning, and has the potential to improve outcomes and pre-operative planning.
Level Of Evidence:
III, case-control retrospective analysis.
