The Skeletal Oncology Research Group Machine Learning Algorithm (SORG-MLA) for predicting prolonged postoperative opioid prescription after total knee arthroplasty: an international validation study using 3,495 patients from a Taiwanese cohort

Abstract

Background Preoperative prediction of prolonged postoperative opioid use (PPOU) after total knee arthroplasty (TKA) could identify high-risk patients for increased surveillance. The Skeletal Oncology Research Group machine learning algorithm (SORG-MLA) has been tested internally while lacking external support to assess its generalizability. The aims of this study were to externally validate this algorithm in an Asian cohort and to identify other potential independent factors for PPOU. Methods In a tertiary center in Taiwan, 3,495 patients receiving TKA from 2010–2018 were included. Baseline characteristics were compared between the external validation cohort and the original developmental cohorts. Discrimination (area under receiver operating characteristic curve [AUROC] and precision-recall curve [AUPRC]), calibration, overall performance (Brier score), and decision curve analysis (DCA) were applied to assess the model performance. A multivariable logistic regression was used to evaluate other potential prognostic factors. Results There were notable differences in baseline characteristics between the validation and the development cohort. Despite these variations, the SORG-MLA (https://sorg-apps.shinyapps.io/tjaopioid/) remained its good discriminatory ability (AUROC, 0.75; AUPRC, 0.34) and good overall performance (Brier score, 0.029; null model Brier score, 0.032). The algorithm could bring clinical benefit in DCA while somewhat overestimating the probability of prolonged opioid use. Preoperative acetaminophen use was an independent factor to predict PPOU (odds ratio, 2.05). Conclusions The SORG-MLA retained its discriminatory ability and good overall performance despite the different pharmaceutical regulations. The algorithm could be used to identify high-risk patients and tailor personalized prevention policy.

Enabling Personalized Medicine in Orthopaedic Surgery Through Artificial Intelligence

“…In addition, models exist to explore the risk of general medical complications after orthopaedic surgery 30,[129][130][131][132][133][134][135][136][137][138][139][140][141][142][143][144][145][146][147] (Table V). Using AI to effectively predict postoperative complications, physicians can attempt to minimize the individual risk of complications through targeted interventions, although this potential has been slow to become realized in clinical practice.…”

“…For institutional resources, many health care systems are partnering with large companies with cloud-based servers and technical AI resources, such as Amazon Web Services. Failure to have established a business relationship to obtain such resources, or lack of technology to facilitate large-scale processes (i.e., 143 THA Identification of patients at high risk of experiencing hyponatremia after THA Buddhiraju et al 130 THA Prediction of blood transfusion after primary and revision THA Ding et al 133 THA Prediction of DVT and PE after THA Shah et al 145 THA Prediction of major complications following THA Chen et al 131 THA Prediction of postoperative delirium after THA Knee Jung et al 138 TKA Prediction of postoperative delirium after TKA Jo et al 137 TKA Development of a web-based predictive model for transfusion after TKA Kolin et al 142 TKA Prediction of risk for postoperative anemia and blood transfusion after TKA Tsai et al 147 TKA Identification of patients at high-risk of prolonged opioid prescription after TKA Klemt et al 141 TKA Prediction of extended opioid use after primary TKA Hip & Knee Shohat et al 146 TKA, THA Prediction of VTE and MBE after TKA and THA Gabriel et al 135 TKA, THA Prediction of persistent opioid use among patients undergoing TKA and THA Onishchenko et al 144 TKA, THA Prediction of major adverse cardiac events after TKA and THA Dai et al 132 TKA, THA Prediction of ischemic stroke risk in patients after TKA and THA Huang et al 136 TKA, THA Prediction of risk for postoperative blood transfusion after TKA and THA Harris et al 30 TKA, THA Prediction of 30-day mortality and cardiac complications after TJA Abraham et al 129 rTKA, rTHA Prediction of 30-day mortality, cardiac complications, and respiratory complications after rTKA and rTHA Spine Kim et al 140 Posterior lumbar spinal fusion Prediction of complications after posterior lumbar spinal fusion Kim et al 139 ASD Prediction of surgical complications after ASD surgery Durand et al 134 ASD Prediction of blood transfusion requirement after ASD surgery *ASD 5 adult spinal deformity, DVT 5 deep venous thrombosis, MBE 5 major bleeding event, PE 5 pulmonary embolism, rTHA 5 revision total hip arthroplasty, rTKA 5 revision total knee arthroplasty, THA 5 total hip arthroplasty, TJA 5 total joint arthroplasty, TKA 5 total knee arthroplasty, and VTE 5 venous thromboembolism.…”

“…This stems from the fact that there are many important quality assurance tasks, such as external validation, that must occur for every AI application before these benefits are seen in a clinical setting. Specifically, many of these studies remain in the derivation 130,132,146 and validation phases 30,138,147 , with no studies implemented in a clinical setting to the best of our knowledge. There are also various results and conclusions among these studies, with some reporting “moderately accurate” predictive models and level III evidence 30 , “excellent performance” with Level IV evidence 141 , and validated and “good performance” among Level II evidence studies 137 .…”