The COVID-19 pandemic created a major backlog of hip and knee replacement procedures across the United Kingdom. To tackle this challenge, surgical hubs have been proposed to boost efficiency, especially for routine, lower-risk operations. These hubs and similar Ambulatory Surgery Centers typically lack advanced critical care support, making them best suited for patients with fewer comorbidities. Accurate preoperative risk assessment is therefore essential to direct patients to the right facility and minimize avoidable complications. This study investigated the application of artificial intelligence (AI) for risk stratification in hip and knee arthroplasty. A polynomial regression model was created using patient demographics, laboratory results, and comorbidities to generate individualized risk scores for postoperative complications. The model was trained on 29,658 patient records drawn from two UK National Health Service (NHS) organizations. Performance was measured using the area under the receiver operating characteristic curve (AUROC), enabling classification of patients into high- and low-risk groups. Validation involved a retrospective review of 445 cases, comparing predicted complications and need for additional care against actual outcomes. The model achieved 70% sensitivity in detecting high-risk patients and a negative predictive value of 96%. Risk predictions from the AI tool proved comparable to those made during traditional consultant-led preoperative evaluations. These results indicate that AI can facilitate more efficient and consistent preoperative risk assessment, helping to ease pressure on assessment teams and improve resource use. Despite certain limitations, the model serves as a valuable support tool for clinical decision-making in determining patient risk. It has potential applications in UK NHS surgical hubs and in Ambulatory Surgery Centers in the United States.
