Artificial Intelligence for the Orthopaedic Surgeon

An Overview of Potential Benefits, Limitations, and Clinical Applications

Eric C. Makhni, MD, MBA; Sonya Makhni, MD, MBA; Prem N. Ramkumar, MD, MBA


J Am Acad Orthop Surg. 2021;29(6):235-243. 

In This Article


The day-to-day employment of ML in an independent orthopaedic practice is not yet widely used. However, individual orthopaedic services across the United States have pioneered the use of ML technology in recent years and may serve as examples of what future deployment will look like. In 2019, Goltz et al,[35] in affiliation with Duke University Medical Center, released a 90-day readmission risk calculator after primary unilateral total hip and knee arthroplasties. Using patient data from 10,155 primary unilateral total hip and knee arthroplasties done at a single institution, a multivariable regression model was created to "adequately predict" the likelihood of 90-day readmission, based on preoperative parameters, duration of surgery, postoperative laboratory results (hemoglobin and blood-urea-nitrogen level), and nine comorbidities. This tool is freely available online for any provider to use and serves as an example of how applied statistical techniques, in conjunction with large amounts of available patient data, can be harnessed to provide benefit in patient care. Although many may not consider such regression models as "ML", it may be useful to consider ML as a natural extension of statistical techniques that have been used in the profession of healthcare for decades. Other online applications are similarly available, ranging from predicting risk of increased length-of-stay after joint arthroplasty to predicting inpatient payments.[21,36,37] It is likely that the use of such applications and online tools, powered by ML algorithms, will become more ubiquitous in the future and paid private consulting services.

In 2018, the Cleveland Clinic's Department of Orthopaedic Surgery established the ML Arthroplasty Laboratory, with the goal of exploring practical implementation of ML techniques in the practice of orthopaedic medicine.[38] The team has developed and validated several ML models in several areas of research interest, all with the ultimate focus of providing patient-specific, value-based care.[20,27,38,39] The team has recently developed an image classifier to read preoperative radiographs and identify arthroplasty implant class and manufacturer before revision. Such a tool would be valuable to any arthroplasty surgeon, avoiding the increase in costs associated with delays in care, and misidentifications leading to lack of appropriate equipment available during the operation.[8] Another area of study for the group has been the establishment of value-based payment models for hip and knee total joint arthroplasty.[22,39] Although the development of these theoretical payment models may not serve any utility for any single orthopaedic practice, these studies represent an initial foray exploring the utility of ML in better informing reimbursement.

Yet a third example of incorporation of AI and ML to routine orthopaedic care lies in with decision-support tools using PROM data to predict outcomes after hip and knee arthroplasties. Using AI, clinicians at the UT Health Austin Musculoskeletal Institute, Dell Medical School at the University of Texas at Austin discuss likelihood of post-operative success before scheduling surgery with patients.[40] These models have the potential to improve in accuracy as more data/inputs are incorporated into the models.

Although the true impact of AI and ML on clinical orthopaedics is still yet to be determined, ample evidence exists that these technologies may assist in generating healthcare value through improving outcomes or decreasing cost/inefficiencies. ML has the capability of automating redundant tasks, thereby allowing physicians to spend more time with patients. The technology should be viewed as a physician-aid—a tool that can better augment a physician's capabilities rather than replace their responsibilities. To maximize the benefit of these tools, however, clinicians, researchers, and policy makers must first understand the fundamentals of the technology, along with its potential benefits and limitations. Numerous applications in orthopaedics have already been demonstrated, and these applications will increase in quantity and impact as AI continues to grow as a key healthcare technology.