Using Presurgical Biopsychosocial Features to Develop an Advanced Clinical Decision-Making Support Tool for Predicting Recovery Trajectories in Patients Undergoing Total Knee Arthroplasty: Protocol for a Prospective Observational Study

Following total knee arthroplasty (TKA), 10% to 20% of patients report dissatisfaction with procedural outcomes. There is growing recognition that postsurgical satisfaction is shaped not only by the quality of surgery but also by psychological and social factors. Surprisingly, information on the psychological and social determinants of surgical outcomes is rarely collected before surgery. A comprehensive collection of biopsychosocial information could assist clinicians in making recommendations in relation to rehabilitation, particularly if there is robust evidence to support the ability of presurgical constructs to predict postsurgical outcomes. Clinical decision support tools can help identify factors influencing patient outcomes and support the provision of interventions or services that can be tailored to meet individuals' needs. However, despite their potential clinical benefit, the application of such tools remains limited. This study aims to develop a clinical decision tool that will assist with patient stratification and more precisely targeted clinical decision-making regarding prehabilitation and rehabilitation for TKA, based on the identified individual biopsychosocial needs. In this prospective observational study, all participants provided written or electronic consent before study commencement. Patient-completed questionnaires captured information related to a broad range of biopsychosocial parameters during the month preceding TKA. These included demographic factors (sex, age, and rurality), psychological factors (mood status, pain catastrophizing, resilience, and committed action), quality of life, social support, lifestyle factors, and knee symptoms. Physical measures assessing mobility, balance, and functional lower body strength were performed via video calls with patients in their home. Information related to preexisting health issues and concomitant medications was derived from hospital medical records. Patient recovery outcomes were assessed 3 months after the surgical procedure and included quality of life, patient-reported knee symptoms, satisfaction with the surgical procedure, and mood status. Machine learning data analysis techniques will be applied to determine which presurgery parameters have the strongest power for predicting patient recovery following total knee replacement. On the basis of these analyses, a predictive model will be developed. Predictive models will undergo internal validation, and Bayesian analysis will be a