Artificial intelligence and machine-learning analytics have gained extensive popularity in recent years due to their clinically relevant applications. A wide range of proof-of-concept studies have demonstrated the ability of these analyses to personalize risk prediction, detect implant specifics from imaging, and monitor and assess patient movement and recovery. Though these applications are exciting and could potentially influence practice, it is imperative to understand when these analyses are indicated and where the data are derived from, prior to investing resources and confidence into the results and conclusions. In this article, we review the current benefits and potential limitations of machine-learning for the orthopaedic surgeon with a specific emphasis on data quality.

Purpose: There is a paucity of literature describing clinical outcomes following hip arthroscopy. Variables associated with short or prolonged recovery are undefined. This presents a challenge to surgeons in preoperatively communicating with patients about expectations after surgery. The goals of this study are to identify predictors of recovery and to develop models which will facilitate the proper counseling of patients prior to hip arthroscopy. In this study, we define a normal recovery after hip arthroscopy, determine the predictive values of preoperative and intraoperative variables for recovery and for progression to total hip arthroplasty (THA) after hip arthroscopy.

Method: A retrospective review of 216 individuals treated with hip arthroscopy at a tertiary medical center was conducted by a single reviewer. Univariate analysis was used to identify independent variables that correlated with prolonged or short recovery following hip arthroscopy and also on variables correlated with progression to THA. Binary logistic regression analysis was used to develop and test multivariate models for predicting prolonged recovery and progression to THA.

Results: Univariate analyses revealed multiple variables (spanning demographics, past medical history, radiographic findings, physical examination findings, and intraoperative findings) which were significantly (p≤0.05) correlated with prolonged recovery (13 significant predictors) and also with progression to THA (14 significant predictors). A multivariate predictive algorithm was generated using 5 significant predictors of prolonged recovery, which included Workman’s compensation involvement, female gender, use of pain medications, presence of a limp, and presence of a lateral labral tear. This algorithm was tested successfully using an independent sample of 25 individuals. Three multivariate predictors of progression to THA after hip arthroscopy were identified, including radiographic presence of arthritis, female gender and the presence of grade 4 chondral lesions, and a predictive algorithm was generated.

Conclusion: We generated and initially validated a multivariate algorithm to predict prolonged recovery following hip arthroscopy. If validated in larger sample, this model may allow a surgeon to appropriately counsel patients regarding expectations for recovery after hip arthroscopy.