The Fundamentals and Applications of Wearable Sensor Devices in Sports Medicine: A Scoping Review.

PURPOSE: To (1) characterize the various forms of wearable sensor devices (WSDs) and (2) review the peer-reviewed literature of applied wearable technology within sports medicine. METHODS: A systematic search of PubMed and EMBASE databases, from inception through 2023, was conducted to identify eligible studies using WSDs within sports medicine. Data extraction was performed of study demographics and sensor specifications. Included studies were categorized by application: athletic training, rehabilitation, and research. RESULTS: In total, 43 studies met criteria for inclusion in this review. Forms of WSDs include pedometers, accelerometers, encoders (consisting of magnetometers and gyroscopes), force sensors, global positioning system trackers, and inertial measurement units. Outcome metrics include step counts; gait, limb motion, and angular positioning; foot and skin pressure; change of direction and inclination, including analysis of both body parts and athletes on a field; displacement and velocity of body segments and joints; heart rate; plethysmography; sport-specific kinematics; range of motion, symmetry, and alignment; head impact; sleep; throwing biomechanics; and kinetic and spatiotemporal running metrics. WSDs are used in athletic training to assess sport-specific biomechanics and workload with a goal of injury prevention and training optimization, as well as for rehabilitation monitoring and research such as for risk predicting and aiding diagnosis. CONCLUSIONS: WSDs enable real-time monitoring of human performance across a variety of implementations and settings, allowing collection of metrics otherwise not achievable. WSDs are powerful tools with multiple applications within athletic training, patient rehabilitation, and orthopaedic and sports medicine research. CLINICAL RELEVANCE: Wearable technology may represent the missing link to quantitatively addressing return to play and previous performance. WSDs are commercially available and portable adjuncts that allow clinicians, trainers, and individual athletes to monitor biomechanical parameters, workload, and recovery status to better contextualize personalized training, injury risk, and rehabilitation.

Who Are the Anatomic Outliers Undergoing Total Knee Arthroplasty? A Computed Tomography-Based Analysis of the Hip-Knee-Ankle Axis Across 1,352 Preoperative Computed Tomographies Using a Deep Learning and Computer Vision-Based Pipeline.

BACKGROUND: Dissatisfaction after total knee arthroplasty (TKA) ranges from 15 to 30%. While patient selection may be partially responsible, morphological and reconstructive challenges may be determinants. Preoperative computed tomography (CT) scans for TKA planning allow us to evaluate the hip-knee-ankle axis and establish a baseline phenotypic distribution across anatomic parameters. The purpose of this cross-sectional analysis was to establish the distributions of 27 parameters in a pre-TKA cohort and perform threshold analysis to identify anatomic outliers. METHODS: There were 1,352 pre-TKA CTs that were processed. A 2-step deep learning pipeline of classification and segmentation models identified landmark images and then generated contour representations. We used an open-source computer vision library to compute measurements for 27 anatomic metrics along the hip-knee axis. Normative distribution plots were established, and thresholds for the 15th percentile at both extremes were calculated. Metrics falling outside the central 70th percentile were considered outlier indices. A threshold analysis of outlier indices against the proportion of the cohort was performed. RESULTS: Significant variation exists in pre-TKA anatomy across 27 normally distributed metrics. Threshold analysis revealed a sigmoid function with a critical point at 9 outlier indices, representing 31.2% of subjects as anatomic outliers. Metrics with the greatest variation related to deformity (tibiofemoral angle, medial proximal tibial angle, lateral distal femoral angle), bony size (tibial width, anteroposterior femoral size, femoral head size, medial femoral condyle size), intraoperative landmarks (posterior tibial slope, transepicondylar and posterior condylar axes), and neglected rotational considerations (acetabular and femoral version, femoral torsion). CONCLUSIONS: In the largest non-industry database of pre-TKA CTs using a fully automated 3-stage deep learning and computer vision-based pipeline, marked anatomic variation exists. In the pursuit of understanding the dissatisfaction rate after TKA, acknowledging that 31% of patients represent anatomic outliers may help us better achieve anatomically personalized TKA, with or without adjunctive technology.

Editorial Commentary: Machine Learning Is Just a Statistical Technique, Not a Mystical Methodology or Peer Review Panacea.

Orthopaedic and sports medicine research surrounding artificial intelligence (AI) has dramatically risen over the last 4 years. Meaningful application and methodologic rigor in the scientific literature are critical to ensure appropriate use of AI. Common but critical errors for those engaging in AI-related research include failure to 1) ensure the question is important and previously unknown or unanswered; 2) establish that AI is necessary to answer the question; and 3) recognize model performance is more commonly a reflection of the data than the AI itself. We must take care to ensure we are not repackaging and internally validating registry data. Instead, we should be critically appraising our data-not the AI-based statistical technique. Without appropriate guardrails surrounding the use of artificial intelligence in Orthopaedic research, there is a risk of repackaging registry data and low-quality research in a recursive peer-reviewed loop.

Patient-Specific Safe Zones for Acetabular Component Positioning in Total Hip Arthroplasty: Mathematically Accounting for Spinopelvic Biomechanics.

BACKGROUND: Despite a growing understanding of spinopelvic biomechanics in total hip arthroplasty (THA), there is no validated approach for executing patient-specific acetabular component positioning. The purpose of this study was to (1) validate quantitative, patient-specific acetabular "safe zone" component positioning from spinopelvic parameters and (2) characterize differences between quantitative patient-specific acetabular targets and qualitative hip-spine classification targets. METHODS: From 2,457 consecutive primary THA patients, 22 (0.88%) underwent revision for instability. Spinopelvic parameters were measured prior to index THA. Acetabular position was measured following index and revision arthroplasty. Using a mathematical proof, we developed an open-source tool translating a surgeon-selected, preoperative standing acetabular target to a patient-specific safe zone intraoperative acetabular target. Difference between the patient-specific safe zone and the actual component position was compared before and after revision. Hip-spine classification targets were compared to patient-specific safe zone targets. RESULTS: Of the 22 who underwent revision, none dislocated at follow-up (4.6 [range, 1 to 6.9]). Patient-specific safe zone targets differed from prerevision acetabular component position by 9.1 ± 4.2° inclination/13.3 ± 6.7° version; after revision, the mean difference was 3.2 ± 3.0° inclination/5.3 ± 2.7° version. Differences between patient-specific safe zones and the median and extremes of recommended hip-spine classification targets were 2.2 ± 1.9° inclination/5.6 ± 3.7° version and 3.0 ± 2.3° inclination/7.9 ± 3.5° version, respectively. CONCLUSION: A mathematically derived, patient-specific approach accommodating spinopelvic biomechanics for acetabular component positioning was validated by approximating revised, now-stable hips within 5° version and 3° inclination. These patient-specific safe zones augment the hip-spine classification with prescriptive quantitative targets for nuanced preoperative planning.

Exposure to Extended Reality and Artificial Intelligence-Based Manifestations: A Primer on the Future of Hip and Knee Arthroplasty.

BACKGROUND: Software-infused services, from robot-assisted and wearable technologies to artificial intelligence (AI)-laden analytics, continue to augment clinical orthopaedics - namely hip and knee arthroplasty. Extended reality (XR) tools, which encompass augmented reality, virtual reality, and mixed reality technology, represent a new frontier for expanding surgical horizons to maximize technical education, expertise, and execution. The purpose of this review is to critically detail and evaluate the recent developments surrounding XR in the field of hip and knee arthroplasty and to address potential future applications as they relate to AI. METHODS: In this narrative review surrounding XR, we discuss (1) definitions, (2) techniques, (3) studies, (4) current applications, and (5) future directions. We highlight XR subsets (augmented reality, virtual reality, and mixed reality) as they relate to AI in the increasingly digitized ecosystem within hip and knee arthroplasty. RESULTS: A narrative review of the XR orthopaedic ecosystem with respect to XR developments is summarized with specific emphasis on hip and knee arthroplasty. The XR as a tool for education, preoperative planning, and surgical execution is discussed with future applications dependent upon AI to potentially obviate the need for robotic assistance and preoperative advanced imaging without sacrificing accuracy. CONCLUSION: In a field where exposure is critical to clinical success, XR represents a novel stand-alone software-infused service that optimizes technical education, execution, and expertise but necessitates integration with AI and previously validated software solutions to offer opportunities that improve surgical precision with or without the use of robotics and computed tomography-based imaging.

Artificial Intelligence for Automated Implant Identification in Total Hip Arthroplasty: A Multicenter External Validation Study Exceeding Two Million Plain Radiographs.

BACKGROUND: The surgical management of complications after total hip arthroplasty (THA) necessitates accurate identification of the femoral implant manufacturer and model. Automated image processing using deep learning has been previously developed and internally validated; however, external validation is necessary prior to responsible application of artificial intelligence (AI)-based technologies. METHODS: We trained, validated, and externally tested a deep learning system to classify femoral-sided THA implants as one of the 8 models from 2 manufacturers derived from 2,954 original, deidentified, retrospectively collected anteroposterior plain radiographs across 3 academic referral centers and 13 surgeons. From these radiographs, 2,117 were used for training, 249 for validation, and 588 for external testing. Augmentation was applied to the training set (n = 2,117,000) to increase model robustness. Performance was evaluated by area under the receiver operating characteristic curve, sensitivity, specificity, and accuracy. Implant identification processing speed was calculated. RESULTS: The training and testing sets were drawn from statistically different populations of implants (P < .001). After 1,000 training epochs by the deep learning system, the system discriminated 8 implant models with a mean area under the receiver operating characteristic curve of 0.991, accuracy of 97.9%, sensitivity of 88.6%, and specificity of 98.9% in the external testing dataset of 588 anteroposterior radiographs. The software classified implants at a mean speed of 0.02 seconds per image. CONCLUSION: An AI-based software demonstrated excellent internal and external validation. Although continued surveillance is necessary with implant library expansion, this software represents responsible and meaningful clinical application of AI with immediate potential to globally scale and assist in preoperative planning prior to revision THA.

Artificial Intelligence for Automated Implant Identification in Knee Arthroplasty: A Multicenter External Validation Study Exceeding 3.5 Million Plain Radiographs.

BACKGROUND: Surgical management of complications following knee arthroplasty demands accurate and timely identification of implant manufacturer and model. Automated image processing using deep machine learning has been previously developed and internally validated; however, external validation is essential prior to scaling clinical implementation for generalizability. METHODS: We trained, validated, and externally tested a deep learning system to classify knee arthroplasty systems as one of the 9 models from 4 manufacturers derived from 4,724 original, retrospectively collected anteroposterior plain knee radiographs across 3 academic referral centers. From these radiographs, 3,568 were used for training, 412 for validation, and 744 for external testing. Augmentation was applied to the training set (n = 3,568,000) to increase model robustness. Performance was determined by the area under the receiver operating characteristic curve, sensitivity, specificity, and accuracy. Implant identification processing speed was calculated. The training and testing sets were drawn from statistically different populations of implants (P < .001). RESULTS: After 1,000 training epochs by the deep learning system, the system discriminated 9 implant models with a mean area under the receiver operating characteristic curve of 0.989, accuracy of 97.4%, sensitivity of 89.2%, and specificity of 99.0% in the external testing dataset of 744 anteroposterior radiographs. The software classified implants at a mean speed of 0.02 seconds per image. CONCLUSION: An artificial intelligence-based software for identifying knee arthroplasty implants demonstrated excellent internal and external validation. Although continued surveillance is necessary with implant library expansion, this software represents a responsible and meaningful clinical application of artificial intelligence with immediate potential to globally scale and assist in preoperative planning prior to revision knee arthroplasty.

Meaningless Applications and Misguided Methodologies in Artificial Intelligence-Related Orthopaedic Research Propagates Hype Over Hope.

There exists great hope and hype in the literature surrounding applications of artificial intelligence (AI) to orthopaedic surgery. Between 2018 and 2021, a total of 178 AI-related articles were published in orthopaedics. However, for every 2 original research papers that apply AI to orthopaedics, a commentary or review is published (30.3%). AI-related research in orthopaedics frequently fails to provide use cases that offer the uninitiated an opportunity to appraise the importance of AI by studying meaningful questions, evaluating unknown hypotheses, or analyzing quality data. The hype perpetuates a feed-forward cycle that relegates AI to a meaningless buzzword by rewarding those with nascent understanding and rudimentary technical knowhow into committing several basic errors: (1) inappropriately conflating vernacular ("AI/machine learning"), (2) repackaging registry data, (3) prematurely releasing internally validated algorithms, (4) overstating the "black box phenomenon" by failing to provide weighted analysis, (5) claiming to evaluate AI rather than the data itself, and (6) withholding full model architecture code. Relevant AI-specific guidelines are forthcoming, but forced application of the original Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis guidelines designed for regression analyses is irrelevant and misleading. To safeguard meaningful use, AI-related research efforts in orthopaedics should be (1) directed toward administrative support over clinical evaluation and management, (2) require the use of the advanced model, and (3) answer a question that was previously unknown, unanswered, or unquantifiable.

Time-Driven Activity-Based Costing Accurately Determines Bundle Cost for Rotator Cuff Repair.

PURPOSE: The purpose of this study was to determine the cost of the episode of care for primary rotator cuff repair (RCR) from day of surgery to 90 days postoperatively using the time-driven activity-based costing (TDABC) method. The secondary purpose of this study was to identify the main drivers of cost for both phases of care. METHODS: This retrospective case series study used the TDABC method to determine the bundled cost of care for an RCR. First, a process map of the RCR episode of care was constructed in order to determine drivers of fixed (i.e., rent, power), direct variable (i.e., healthcare personnel), and indirect costs (i.e., marketing, building maintenance). The study was performed at a Midwestern tertiary care medical system, and patients were included in the study if they underwent an RCR from January 2018 to January 2019 with at least 90 days of postoperative follow-up. In this article, all costs were included, but we did not account for fees to provider and professional groups. RESULTS: The TDABC method calculated a cost of $10,569 for a bundled RCR, with 76% arising from the operative phase and 24% from the postoperative phase. The main driver of cost within the operative phase was the direct fixed costs, which accounted for 35% of the cost in this phase, and the largest contributor to cost within this category was the cost of implants, which accounted for 55%. In the postoperative phase of care, physical therapy visits were the greatest contributor to cost at 59%. CONCLUSION: In a bundled cost of care for RCR, the largest cost driver occurs on the day of surgery for direct fixed costs, in particular, the implant. Physical therapy represents over half of the costs of the episode of care. Better understanding the specific cost of care for RCR will facilitate optimization with appropriately designed payment models and policies that safeguard the interests of the patient, physician, and payer. LEVEL OF EVIDENCE: IV, therapeutic case series.

Preoperative Magnetic Resonance Imaging Offers Questionable Clinical Utility, Delays Time to Hip Arthroscopy, and Lacks Cost-Effectiveness in Patients Aged ≤40 Years With Femoroacetabular Impingement Syndrome: A Retrospective 5-Year Analysis.

PURPOSE: To assess the clinical utility of preoperative magnetic resonance imaging (MRI) and quantify the delay in surgical care for patients aged ≤40 years undergoing primary hip arthroscopy with history, physical examination, and radiographs concordant with femoroacetabular impingement syndrome (FAIS). METHODS: From August 2015 to December 2020, 1,786 consecutive patients were reviewed from the practice of 1 fellowship-trained hip arthroscopist. Inclusion criteria were FAIS, primary surgery, and age ≤40 years. Exclusion criteria were MRI contraindication, reattempt of conservative management, or concomitant periacetabular osteotomy. After nonoperative treatment options were exhausted and a surgical plan was established, patients were stratified by those who presented with versus without MRI. Those without existing MRI received one, and any deviations from the surgical plan were noted. All preoperative MRIs were compared with office evaluation and intraoperative findings to assess agreement. Demographic data, Hip Disability and Osteoarthritis Outcome Score (HOOS)-Pain, and time from office to MRI or arthroscopy were recorded. RESULTS: Of the patients indicated by history, physical examination, and radiographs alone (70% female, body mass index 24.8 kg/m2, age 25.9 years), 198 patients presented without MRI and 934 with MRI. None of the 198 had surgical plans altered after MRI. Patients in both groups had MRI findings demonstrating anterosuperior labral tears that were visualized and repaired intraoperatively. Mean time from office to arthroscopy for patients without MRI versus those with was 107.0 ± 67 and 85.0 ± 53 days, respectively (P < .001). Time to MRI was 22.8 days. No difference between groups was observed among the 85% of patients who surpassed the HOOS-Pain minimal clinically important difference (MCID). CONCLUSION: Once indicated for surgery based on history, physical examination, and radiographs, preoperative MRI did not alter the surgical plan for patients aged ≤40 years with FAIS undergoing primary hip arthroscopy. Moreover, preoperative MRI delayed time to arthroscopy. The necessity of routine preoperative MRI in the young primary FAIS population should be challenged.

The development and deployment of machine learning models.

Applications of artificial intelligence, specifically machine learning, are becoming increasingly popular in Orthopaedic Surgery, and medicine as a whole. This growing interest is shared by data scientists and physicians alike. However, there is an asymmetry of understanding of the developmental process and potential applications of machine learning. As new technology will undoubtedly affect clinical practice in the coming years, it is important for physicians to understand how these processes work. The purpose of this paper is to provide clarity and a general framework for building and assessing machine learning models.

Pectoralis muscle injuries in Major and Minor League Baseball.

BACKGROUND AND HYPOTHESIS: Although shoulder and elbow injuries in professional baseball players have been thoroughly studied, little is known about the frequency and impact of pectoralis muscle injuries in this population. The purpose of this study was to use the official league injury surveillance system to describe pectoralis muscle injuries in professional baseball players in Major League Baseball (MLB) and Minor League Baseball (MiLB). Specifically, (1) player demographic characteristics, (2) return to play (RTP), (3) injury mechanism, (4) throwing- and batting-side dominance, and (5) injury rate per athlete exposure (AE) were characterized to guide future injury prevention strategies. METHODS: The MLB Health and Injury Tracking System database was used to compile all pectoralis muscle injuries in MLB and MiLB athletes in the 2011-2017 seasons. Injury-related data including diagnosis (tear or rupture vs. strain), player demographic characteristics, injury timing, need for surgical intervention, RTP, and mechanism of injury were recorded. Subanalyses of throwing- and batting-side dominance, as well as MLB vs. MiLB injury frequency, were performed. RESULTS: A total of 138 pectoralis muscle injuries (32 MLB and 106 MiLB injuries) were reported in the study period (5 tears or ruptures and 133 strains), with 5 of these being recurrent injuries. Operative intervention was performed in 4 athletes (2.9%). Of the 138 injuries, 116 (84.1%) resulted in missed days of play, with a mean time to RTP of 19.5 days. Starting pitchers sustained the greatest proportion of pectoralis injuries (48.1%), with pitching being the most common activity at the time of injury (45.9%). A majority of injuries (86.5%) were sustained during non-contact play. Overall, 87.5% of injuries occurred on the player's dominant throwing side and 81.3% occurred on the player's dominant batting side. There was no significant difference in the rate of pectoralis injuries in the MLB regular season (0.584 per 10,000 AEs) vs. the MiLB regular season (0.425 per 10,000 AEs) (P = .1018). CONCLUSION: Pectoralis muscle injuries are most frequently non-contact injuries, most commonly sustained by pitchers. An understanding of these injuries can guide athletic trainers and management in expectation management and decision making, in addition to directing future efforts at injury prevention.

Psoriasis and Post-Surgical Infections in Primary Total Knee Arthroplasty: An Analysis of 10,727 Patients.

BACKGROUND: Psoriasis is a dermatologic condition characterized by erythematous plaques that may increase wound complications and deep infections following total knee arthroplasty (TKA). There is a paucity of evidence concerning the association of this disease and complications after TKA. This study aimed to determine if patients who have psoriasis vs non-psoriatic patients have differences in demographics and various comorbidities as well as post-operative infections, specifically the following: (1) wound complications; (2) cellulitic episodes; and (3) deep surgical site infections (SSIs). METHODS: We identified 10,727 patients undergoing primary TKA utilizing an institutional database between January 1, 2017 and April 1, 2019. A total of 133 patients who had psoriasis (1.2%) were identified using International Classification of Diseases, Tenth Revision codes and compared to non-psoriatic patients. The rate of wound complications, cellulitic episodes, and deep SSIs were determined. After controlling for age and various comorbidities, multivariate analyses were performed to identify the associated risks for post-operative infections. RESULTS: Psoriasis patients showed an increased associated risk of deep SSIs (3.8%) compared to non-psoriasis patients (1.2%, P = .023). Multivariate analyses demonstrated a significant associated risk of deep SSIs (odds ratio 7.04, 95% confidence interval 2.38-20.9, P < .001) and wound complications (odds ratio 4.44, 95% confidence interval 1.02-19.2, P = .047). CONCLUSION: Psoriasis is an inflammatory dermatologic condition that warrants increased pre-operative counseling, shared decision-making, and infectious precautions in the TKA population given the increased risk of wound complications and deep SSIs. Increased vigilance is required given the coexistence of certain comorbidities with this population, including depression, substance use disorder, smoking history, gastroesophageal reflux disease, and inflammatory bowel disease.

Clinical and Research Medical Applications of Artificial Intelligence.

Artificial intelligence (AI), including machine learning (ML), has transformed numerous industries through newfound efficiencies and supportive decision-making. With the exponential growth of computing power and large datasets, AI has transitioned from theory to reality in teaching machines to automate tasks without human supervision. AI-based computational algorithms analyze "training sets" using pattern recognition and learning from inputted data to classify and predict outputs that otherwise could not be effectively analyzed with human processing or standard statistical methods. Though widespread understanding of the fundamental principles and adoption of applications have yet to be achieved, recent applications and research efforts implementing AI have demonstrated great promise in predicting future injury risk, interpreting advanced imaging, evaluating patient-reported outcomes, reporting value-based metrics, and augmenting telehealth. With appreciation, caution, and experience applying AI, the potential to automate tasks and improve data-driven insights may be realized to fundamentally improve patient care. The purpose of this review is to discuss the pearls, pitfalls, and applications associated with AI.

Epidemiology of acromioclavicular joint injuries in professional baseball: analysis from the Major League Baseball Health and Injury Tracking System.

BACKGROUND: Shoulder injuries account for a large portion of all recorded injuries in professional baseball. Much is known about other shoulder pathologies in the overhead athlete, but the incidence and impact of acromioclavicular (AC) joint injuries in this population are unknown. We examined the epidemiology of AC joint injuries in Major League Baseball (MLB) and Minor League Baseball (MiLB) players and determined the impact on time missed. METHODS: The MLB Health and Injury Tracking System was used to compile records of all MLB and MiLB players from 2011 to 2017 with documented AC joint injuries. These injuries were classified as acute (sprain or separation) or chronic (AC joint arthritis or distal clavicular osteolysis), and associated data extracted included laterality, date of injury, player position, activity, mechanism of injury, length of return to play, and need for surgical intervention. RESULTS: A total of 312 AC joint injuries (183 in MiLB players and 129 in MLB players; range, 39-60 per year) were recorded: 201 acute (64.4%) and 111 chronic (35.6%). A total of 81% of acute and 59% of chronic injuries resulted in time missed, with a mean length of return to play of 21 days for both. Of the injuries in outfielders, 79.6% were acute (P < .0001), as were 66.3% of injuries in infielders (P = .004). Pitchers and catchers had more equal proportions of acute and chronic AC injuries (P > .05 for all). Acute AC injuries occurred most often while fielding (n = 100, 84.7%), running (n = 25, 80.6%), and hitting (n = 19, 61.3%), whereas chronic injuries tended to be more common while pitching (n = 26, 68.4%). Of contact injuries, 82.5% were acute (P < .0001), whereas 59.0% of noncontact injuries were chronic (P = .047). MLB players showed consistently higher regular-season rates of both acute and chronic AC injuries than MiLB players (P < .0001 for each). CONCLUSION: Acute AC joint injuries are contact injuries occurring most commonly among infielders and outfielders while fielding that result in 3 weeks missed before return to play, whereas chronic AC joint injuries occur more commonly in pitchers and catchers from noncontact repetitive overhead activity. Knowledge of these data can better guide expectation management in this elite population to better elucidate the prevalence of 2 common injury patterns in the AC joint.

Artificial Intelligence to Identify Arthroplasty Implants From Radiographs of the Knee.

BACKGROUND: Revisions and reoperations for patients who have undergone total knee arthroplasty (TKA), unicompartmental knee arthroplasty (UKA), and distal femoral replacement (DFR) necessitates accurate identification of implant manufacturer and model. Failure risks delays in care, increased morbidity, and further financial burden. Deep learning permits automated image processing to mitigate the challenges behind expeditious, cost-effective preoperative planning. Our aim was to investigate whether a deep-learning algorithm could accurately identify the manufacturer and model of arthroplasty implants about the knee from plain radiographs. METHODS: We trained, validated, and externally tested a deep-learning algorithm to classify knee arthroplasty implants from one of 9 different implant models from retrospectively collected anterior-posterior (AP) plain radiographs from four sites in one quaternary referral health system. The performance was evaluated by calculating the area under the receiver-operating characteristic curve (AUC), sensitivity, specificity, and accuracy when compared with a reference standard of implant model from operative reports. RESULTS: The training and validation data sets were comprised of 682 radiographs across 424 patients and included a wide range of TKAs from the four leading implant manufacturers. After 1000 training epochs by the deep-learning algorithm, the model discriminated nine implant models with an AUC of 0.99, accuracy 99%, sensitivity of 95%, and specificity of 99% in the external-testing data set of 74 radiographs. CONCLUSIONS: A deep learning algorithm using plain radiographs differentiated between 9 unique knee arthroplasty implants from four manufacturers with near-perfect accuracy. The iterative capability of the algorithm allows for scalable expansion of implant discriminations and represents an opportunity in delivering cost-effective care for revision arthroplasty.

Artificial Intelligence to Identify Arthroplasty Implants From Radiographs of the Hip.

BACKGROUND: The surgical management of complications surrounding patients who have undergone hip arthroplasty necessitates accurate identification of the femoral implant manufacturer and model. Failure to do so risks delays in care, increased morbidity, and further economic burden. Because few arthroplasty experts can confidently classify implants using plain radiographs, automated image processing using deep learning for implant identification may offer an opportunity to improve the value of care rendered. METHODS: We trained, validated, and externally tested a deep-learning system to classify total hip arthroplasty and hip resurfacing arthroplasty femoral implants as one of 18 different manufacturer models from 1972 retrospectively collected anterior-posterior (AP) plain radiographs from 4 sites in one quaternary referral health system. From these radiographs, 1559 were used for training, 207 for validation, and 206 for external testing. Performance was evaluated by calculating the area under the receiver-operating characteristic curve, sensitivity, specificity, and accuracy, as compared with a reference standard of implant model from operative reports with implant serial numbers. RESULTS: The training and validation data sets from 1715 patients and 1766 AP radiographs included 18 different femoral components across four leading implant manufacturers and 10 fellowship-trained arthroplasty surgeons. After 1000 training epochs by the deep-learning system, the system discriminated 18 implant models with an area under the receiver-operating characteristic curve of 0.999, accuracy of 99.6%, sensitivity of 94.3%, and specificity of 99.8% in the external-testing data set of 206 AP radiographs. CONCLUSIONS: A deep-learning system using AP plain radiographs accurately differentiated among 18 hip arthroplasty models from four industry leading manufacturers.

The value of artificial neural networks for predicting length of stay, discharge disposition, and inpatient costs after anatomic and reverse shoulder arthroplasty.

HYPOTHESIS/PURPOSE: The objective is to develop and validate an artificial intelligence model, specifically an artificial neural network (ANN), to predict length of stay (LOS), discharge disposition, and inpatient charges for primary anatomic total (aTSA), reverse total (rTSA), and hemi- (HSA) shoulder arthroplasty to establish internal validity in predicting patient-specific value metrics. METHODS: Using data from the National Inpatient Sample between 2003 and 2014, 4 different ANN models to predict LOS, discharge disposition, and inpatient costs using 39 preoperative variables were developed based on diagnosis and arthroplasty type: primary chronic/degenerative aTSA, primary chronic/degenerative rTSA, primary traumatic/acute rTSA, and primary acute/traumatic HSA. Models were also combined into diagnosis type only. Outcome metrics included accuracy and area under the curve (AUC) for a receiver operating characteristic curve. RESULTS: A total of 111,147 patients undergoing primary shoulder replacement were included. The machine learning algorithm predicting the overall chronic/degenerative conditions model (aTSA, rTSA) achieved accuracies of 76.5%, 91.8%, and 73.1% for total cost, LOS, and disposition, respectively; AUCs were 0.75, 0.89, and 0.77 for total cost, LOS, and disposition, respectively. The overall acute/traumatic conditions model (rTSA, HSA) had accuracies of 70.3%, 79.1%, and 72.0% and AUCs of 0.72, 0.78, and 0.79 for total cost, LOS, and discharge disposition, respectively. CONCLUSION: Our ANN demonstrated fair to good accuracy and reliability for predicting inpatient cost, LOS, and discharge disposition in shoulder arthroplasty for both chronic/degenerative and acute/traumatic conditions. Machine learning has the potential to preoperatively predict costs, LOS, and disposition using patient-specific data for expectation management between health care providers, patients, and payers.

Prognostic utility of an magnetic resonance imaging-based classification for operative versus nonoperative management of ulnar collateral ligament tears: one-year follow-up.

BACKGROUND: A recently introduced classification of medial ulnar collateral ligament (UCL) tears has demonstrated high interobserver and intraobserver reliability, but little is known about its prognostic utility. The purpose of this study was to assess the relationship of the magnetic resonance imaging (MRI)-based classification system and nonoperative vs. operative management. Secondary objectives included subanalysis of baseball players. METHODS: Eighty-five consecutive patients with UCL tears after a standardized treatment paradigm were categorized as operative vs. nonoperative. UCL tears of patients with a minimum of 1-year follow-up were retrospectively classified using the MRI-based classification system. Subanalyses for baseball players included return-to-play and return-to-prior performance. RESULTS: A total of 80 patients (62 baseball players, 54 pitchers) met inclusion criteria. A total of 51 patients underwent surgery, and 29 patients completed nonoperative management. In baseball players, 59% of the proximal tears were treated nonoperatively and 97% of the distal tears were treated operatively; 100% of the proximal partial-thickness tears and 100% of the distal complete tears were treated nonoperatively and operatively, respectively. Patients with distal (odds ratio: 48.4, P < .0001) and complete (odds ratio: 5.0, P = .004) tears were more likely to undergo surgery. Baseball players, regardless of position, were determinants of operative management, and there was no difference in return-to-play clearance and return-to-prior performance between the operative and nonoperative groups. CONCLUSION: A reliable 6-stage MRI-based classification addressing UCL tear grade and location may confer decision making between operative and nonoperative management. Complete and distal tears carry a markedly increased risk of failing nonoperative care compared with proximal, partial tears.

Artificial Intelligence and Arthroplasty at a Single Institution: Real-World Applications of Machine Learning to Big Data, Value-Based Care, Mobile Health, and Remote Patient Monitoring.

BACKGROUND: Driven by the recent ubiquity of big data and computing power, we established the Machine Learning Arthroplasty Laboratory (MLAL) to examine and apply artificial intelligence (AI) to musculoskeletal medicine. METHODS: In this review, we discuss the 2 core objectives of the MLAL as they relate to the practice and progress of orthopedic surgery: (1) patient-specific, value-based care and (2) human movement. RESULTS: We developed and validated several machine learning-based models for primary lower extremity arthroplasty that preoperatively predict patient-specific, risk-adjusted value metrics, including cost, length of stay, and discharge disposition, to provide improved expectation management, preoperative planning, and potential financial arbitration. Additionally, we leveraged passive, ubiquitous mobile technologies to build a small data registry of human movement surrounding TKA that permits remote patient monitoring to evaluate therapy compliance, outcomes, opioid intake, mobility, and joint range of motion. CONCLUSION: The rapid rate with which we in arthroplasty are acquiring and storing continuous data, whether passively or actively, demands an advanced processing approach: AI. By carefully studying AI techniques with the MLAL, we have applied this evolving technique as a first step that may directly improve patient outcomes and practice of orthopedics.