Comparative Analysis of the Ability of Machine Learning Models in Predicting In-hospital Postoperative Outcomes After Total Hip Arthroplasty

Abstract

Background: Machine learning (ML) methods have shown promise in a wide range of applications including the development of patientspecific predictive models before surgical interventions. The purpose of this study was to develop, test, and compare four distinct ML models to predict postoperative parameters after primary total hip arthroplasty. Methods: Data from the Nationwide Inpatient Sample were used to identify patients undergoing total hip arthroplasty from 2016 to 2017. Linear support vector machine (LSVM), random forest (RF), neural network (NN), and extreme gradient boost trees (XGBoost) predictive of mortality, length of stay, and discharge disposition were developed and validated using 15 predictive patient-specific and hospital-specific factors. Area under the curve of the receiver operating characteristic (AUCROC) curve and accuracy were used as validity metrics, and the strongest predictive variables under each model were assessed. Results: A total of 177,442 patients were included in this analysis. For mortality, the XGBoost, NN, and LSVM models all had excellent responsiveness during validation while RF had fair responsiveness. LSVM had the highest responsiveness with an AUCROC of 0.973 during validation. For the length of stay, the LSVM and NN models had fair responsiveness while the XGBoost and random forest models had poor responsiveness. LSVM had the highest responsiveness with an AUCROC of 0.744 during validation. For the discharge disposition outcome, LSVM had good responsiveness while the XGBoost, NN, and RF models all had fair responsiveness. LSVM had the highest responsiveness with an AUCROC of 0.801. Discussion: The ML methods tested demonstrated a range of poorto-excellent responsiveness and accuracy in the prediction of the assessed metrics, with LSVM being the best performer. Such models

Optimizing Total Hip and Knee Arthroplasty Among an Underserved Population: Lessons Learned From a Quality-Improvement Initiative

“…There has been a recent push for the implementation of patient-specific reimbursement models to avoid any unintended limitation to access-to-care that could ensue from an outcome-based reimbursement system, especially among minorities and patients with higher risk of complications [ 18 , 19 ]. Proponents of such models argue that, within the current system, surgeons would limit offering procedures to patients with risk-factors known to be associated with a higher rate of complication or suboptimal outcomes.…”

Epidemiology of total hip arthroplasty: demographics, comorbidities and outcomes

“…More recently, the implementation of advanced data analysis tools, such as machine learning algorithms, to better understand impact of a combination of various medical comorbidities on postoperative outcomes has been popularized [28][29][30]. Harris et al demonstrated an accurate predictive model for mortality and complications following joint arthroplasty with patient-specific variables [31].…”

Epidemiology of primary and revision total knee arthroplasty: analysis of demographics, comorbidities and outcomes from the national inpatient sample

“…Recently, machine learning algorithms have been gaining popularity in the prediction of postoperative outcomes after TKA [ 24 , 25 ]. Comorbidities can be incorporated into such algorithms to develop accurate models of patient risk stratification.…”