The use of artificial intelligence (AI) is rapidly growing across many domains, of which the medical field is no exception. AI is an umbrella term defining the practical application of algorithms to generate useful output, without the need of human cognition. Owing to the expanding volume of patient information collected, known as ‘big data’, AI is showing promise as a useful tool in healthcare research and across all aspects of patient care pathways. Practical applications in orthopaedic surgery include: diagnostics, such as fracture recognition and tumour detection; predictive models of clinical and patient-reported outcome measures, such as calculating mortality rates and length of hospital stay; and real-time rehabilitation monitoring and surgical training. However, clinicians should remain cognizant of AI’s limitations, as the development of robust reporting and validation frameworks is of paramount importance to prevent avoidable errors and biases. The aim of this review article is to provide a comprehensive understanding of AI and its subfields, as well as to delineate its existing clinical applications in trauma and orthopaedic surgery. Furthermore, this narrative review expands upon the limitations of AI and future direction. Cite this article: Bone Joint Res 2023;12(7):447–454

Aims. The principles of evidence-based medicine (EBM) are the foundation of modern medical practice. Surgeons are familiar with the commonly used statistical techniques to test hypotheses, summarize findings, and provide answers within a specified range of probability. Based on this knowledge, they are able to critically evaluate research before deciding whether or not to adopt the findings into practice. Recently, there has been an increased use of artificial intelligence (AI) to analyze information and derive findings in orthopaedic research. These techniques use a set of statistical tools that are increasingly complex and may be unfamiliar to the orthopaedic surgeon. It is unclear if this shift towards less familiar techniques is widely accepted in the orthopaedic community. This study aimed to provide an exploration of understanding and acceptance of AI use in research among orthopaedic surgeons. Methods. Semi-structured in-depth interviews were carried out on a sample of 12 orthopaedic surgeons. Inductive thematic analysis was used to identify key themes. Results. The four intersecting themes identified were: 1) validity in traditional research, 2) confusion around the definition of AI, 3) an inability to validate AI research, and 4) cautious optimism about AI research. Underpinning these themes is the notion of a validity heuristic that is strongly rooted in traditional research teaching and embedded in medical and surgical training. Conclusion. Research involving AI sometimes challenges the accepted traditional evidence-based framework. This can give rise to confusion among orthopaedic surgeons, who may be unable to confidently validate findings. In our study, the impact of this was mediated by cautious optimism based on an ingrained validity heuristic that orthopaedic surgeons develop through their medical training. Adding to this, the integration of AI into everyday life works to reduce suspicion and aid acceptance. Cite this article: Bone Jt Open 2023;4(9):696–703

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

Aims. This study was designed to develop a model for predicting bone mineral density (BMD) loss of the femur after total hip arthroplasty (THA) using artificial intelligence (AI), and to identify factors that influence the prediction. Additionally, we virtually examined the efficacy of administration of bisphosphonate for cases with severe BMD loss based on the predictive model. Methods. The study included 538 joints that underwent primary THA. The patients were divided into groups using unsupervised time series clustering for five-year BMD loss of Gruen zone 7 postoperatively, and a machine-learning model to predict the BMD loss was developed. Additionally, the predictor for BMD loss was extracted using SHapley Additive exPlanations (SHAP). The patient-specific efficacy of bisphosphonate, which is the most important categorical predictor for BMD loss, was examined by calculating the change in predictive probability when hypothetically switching between the inclusion and exclusion of bisphosphonate. Results. Time series clustering allowed us to divide the patients into two groups, and the predictive factors were identified including patient- and operation-related factors. The area under the receiver operating characteristic (ROC) curve (AUC) for the BMD loss prediction averaged 0.734. Virtual administration of bisphosphonate showed on average 14% efficacy in preventing BMD loss of zone 7. Additionally, stem types and preoperative triglyceride (TG), creatinine (Cr), estimated glomerular filtration rate (eGFR), and creatine kinase (CK) showed significant association with the estimated patient-specific efficacy of bisphosphonate. Conclusion. Periprosthetic BMD loss after THA is predictable based on patient- and operation-related factors, and optimal prescription of bisphosphonate based on the prediction may prevent BMD loss. Cite this article: Bone Joint Res 2024;13(4):184–192

Cite this article: Bone Joint Res 2023;12(8):494–496.

In the last decades, the use of artificial intelligence (AI) has been increasingly investigated in intervertebral disc degeneration (IDD) and chronic low back pain (LBP) research. To date, several AI-based cutting-edge technologies, such as computer vision, computer-assisted diagnosis, decision support system and natural language processing have been utilized to optimize LBP prevention, diagnosis, and treatment. This talk will provide an outline on contemporary AI applications to IDD and LBP research, with a particular attention towards actual knowledge gaps and promising innovative tools

Over 8000 total hip arthroplasties (THA) in the UK were revised in 2019, half for aseptic loosening. It is believed that Artificial Intelligence (AI) could identify or predict failing THA and result in early recognition of poorly performing implants and reduce patient suffering. The aim of this study is to investigate whether Artificial Intelligence based machine learning (ML) / Deep Learning (DL) techniques can train an algorithm to identify and/or predict failing uncemented THA. Consent was sought from patients followed up in a single design, uncemented THA implant surveillance study (2010–2021). Oxford hip scores and radiographs were collected at yearly intervals. Radiographs were analysed by 3 observers for presence of markers of implant loosening/failure: periprosthetic lucency, cortical hypertrophy, and pedestal formation. DL using the RGB ResNet 18 model, with images entered chronologically, was trained according to revision status and radiographic features. Data augmentation and cross validation were used to increase the available training data, reduce bias, and improve verification of results. 184 patients consented to inclusion. 6 (3.2%) patients were revised for aseptic loosening. 2097 radiographs were analysed: 21 (11.4%) patients had three radiographic features of failure. 166 patients were used for ML algorithm testing of 3 scenarios to detect those who were revised. 1) The use of revision as an end point was associated with increased variability in accuracy. The area under the curve (AUC) was 23–97%. 2) Using 2/3 radiographic features associated with failure was associated with improved results, AUC: 75–100%. 3) Using 3/3 radiographic features, had less variability, reduced AUC of 73%, but 5/6 patients who had been revised were identified (total 66 identified). The best algorithm identified the greatest number of revised hips (5/6), predicting failure 2–8 years before revision, before all radiographic features were visible and before a significant fall in the Oxford Hip score. True-Positive: 0.77, False Positive: 0.29. ML algorithms can identify failing THA before visible features on radiographs or before PROM scores deteriorate. This is an important finding that could identify failing THA early

Bone age is a radiographical assessment used in pediatric medicine due to its relative objectivity in determining biological maturity compared to chronological age and size.1 Currently, Greulich and Pyle (GP) is one of the most common methods used to determine bone age from hand radiographs.2–4 In recent years, new methods were developed to increase the efficiency in bone age analysis like the shorthand bone age (SBA) and the automated artificial intelligence algorithms. The purpose of this study is to evaluate the accuracy and reliability of these two methods and examine if the reduction in analysis time compromises their accuracy. Two hundred thirteen males and 213 females were selected. Each participant had their bone age determined by two separate raters using the GP (M1) and SBA methods (M2). Three weeks later, the two raters repeated the analysis of the radiographs. The raters timed themselves using an online stopwatch while analyzing the radiograph on a computer screen. De-identified radiographs were securely uploaded to an automated algorithm developed by a group of radiologists in Toronto. The gold standard was determined to be the radiology report attached to each radiograph, written by experienced radiologists using GP (M1). For intra-rater variability, intraclass correlation analysis between trial 1 (T1) and trial 2 (T2) for each rater and method was performed. For inter-rater variability, intraclass correlation was performed between rater 1 (R1) and rater 2 (R2) for each method and trial. Intraclass correlation between each method and the gold standard fell within the 0.8–0.9 range, highlighting significant agreement. Most of the comparisons showed a statistically significant difference between the two new methods and the gold standard; however it may not be clinically significant as it ranges between 0.25–0.5 years. A bone age is considered clinically abnormal if it falls outside 2 standard deviations of the chronological age; standard deviations are calculated and provided in GP atlas.6–8 For a 10-year old female, 2 standard deviations constitute 21.6 months which far outweighs the difference reported here between SBA, automated algorithm and the gold standard. The median time for completion using the GP method was 21.83 seconds for rater 1 and 9.30 seconds for rater 2. In comparison, SBA required a median time of 7 seconds for rater 1 and 5 seconds for rater 2. The automated method had no time restraint as bone age was determined immediately upon radiograph upload. The correlation between the two trials in each method and rater (i.e. R1M1T1 vs R1M1T2) was excellent (κ= 0.9–1) confirming the reliability of the two new methods. Similarly, the correlation between the two raters in each method and trial (i.e. R1M1T1 vs R2M1T1) fell within the 0.9–1 range. This indicates a limited variability between raters who may use these two methods. The shorthand bone age method and an artificial intelligence automated algorithm produced values that are in agreement with the gold standard Greulich and Pyle, while reducing analysis time and maintaining a high inter-rater and intra-rater reliability

As patient data continues to grow, the importance of efficient and precise analysis cannot be overstated. The employment of Generative Artificial Intelligence (AI), specifically Chat GPT-4, in the realm of medical data interpretation has been on the rise. However, its effectiveness in comparison to manual data analysis has been insufficiently investigated. This quality improvement project aimed to evaluate the accuracy and time-efficiency of Generative AI (GPT-4) against manual data interpretation within extensive datasets pertaining to patients with orthopaedic injuries. A dataset, containing details of 6,562 orthopaedic trauma patients admitted to a district general hospital over a span of two years, was reviewed. Two researchers operated independently: one utilised GPT-4 for insights via prompts, while the other manually examined the identical dataset employing Microsoft Excel and IBM® SPSS® software. Both were blinded on each other's procedures and outcomes. Each researcher answered 20 questions based on the dataset including injury details, age groups, injury specifics, activity trends and the duration taken to assess the data. Upon comparison, both GPT-4 and the manual researcher achieved consistent results for 19 out of the 20 questions (95% accuracy). After a subsequent review and refined prompts (prompt engineering) to GPT-4, the answer to the final question aligned with the manual researcher's findings. GPT-4 required just 30 minutes, a stark contrast to the manual researcher's 9-hour analytical duration. This quality improvement project emphasises the transformative potential of Generative AI in the domain of medical data analysis. GPT-4 not only paralleled the accuracy of manual analysis but also achieved this in significantly less time. For optimal accurate results, data analysis by AI can be enhanced through human oversight. Adopting AI-driven approaches, particularly in orthopaedic data interpretation, can enhance efficiency and ultimately improve patient care. We recommend future investigations on large and more varied datasets to reaffirm these outcomes

INTRODUCTION. Quality monitoring is increasingly important to support and assure sustainability of the Orthopaedic practice. Many surgeons in a non-academic setting lack the resources to accurately monitor quality of care. Widespread use of electronic medical records (EMR) provides easier access to medical information and facilitates its analysis. However, manual review of EMRs is inefficient and costly. Artificial Intelligence (AI) software has allowed for development of automated search algorithms for extracting relevant complications from EMRs. We questioned whether an AI supported algorithm could be used to provide accurate feedback on the quality of care following Total Hip Arthroplasty (THA) in a high-volume, non-academic setting. METHODS. 532 Consecutive patients underwent 613 THA between January 1. st. and December 31. st. , 2017. Patients were prospectively followed pre-op, 6 weeks, 3 months and 1 year. They were seen by the surgeon who created clinical notes and reported every adverse event. A random derivation cohort (100 patients, 115 hips) was used to determine accuracy. The algorithm was compared to manual extraction to validate performance in raw data extraction. The full cohort (532 patients, 613 hips) was used to determine its recall, precision and F-value. RESULTS. The algorithm had an accuracy value of 95.0%, compared to 94.5% for manual review (p=0.69). Recall of 96.0% was achieved with precision of 88.0% and F-measure of 0.85 for all adverse events. Recovery of 80.6% of patients was completely uneventful. Re-intervention was required in 1.3% of cases and 18.1% had a ‘transient’ event such as low back pain. The infection and dislocation rate was 0,3%. CONCLUSION. An AI supported search algorithm can analyze and interpret large quantities of EMRs at greater speed but with performance comparable to manual review. Using the program, new clinical information surfaced. 18.1% of patients can be expected to have a ‘transient’ problem following a THA procedure

Introduction and Objective

Up to 30% of thoracolumbar (TL) fractures are missed in the emergency room. Failure to identify these fractures can result in neurological injuries up to 51% of the casesthis article aimed to clarify the incidence and risk factors of traumatic fractures in China. The China National Fracture Study (CNFS. Obtaining sagittal and anteroposterior radiographs of the TL spine are the first diagnostic step when suspecting a traumatic injury. In most cases, CT and/or MRI are needed to confirm the diagnosis. These are time and resource consuming. Thus, reliably detecting vertebral fractures in simple radiographic projections would have a significant impact. We aim to develop and validate a deep learning tool capable of detecting TL fractures on lateral radiographs of the spine. The clinical implementation of this tool is anticipated to reduce the rate of missed vertebral fractures in emergency rooms.

Polyethylene wear represents a significant risk factor for the long-term success of knee arthroplasty [1]. This work aimed to develop and in vivo validate an automated algorithm for accurate and precise AI based wear measurement in knee arthroplasty using clinical AP radiographs for scientifically meaningful multi-centre studies.

Twenty postoperative radiographs (knee joint AP in standing position) after knee arthroplasty were analysed using the novel algorithm. A convolutional neural network-based segmentation is used to localize the implant components on the X-Ray, and a 2D-3D registration of the CAD implant models precisely calculates the three-dimensional position and orientation of the implants in the joint at the time of acquisition. From this, the minimal distance between the involved implant components is determined, and its postoperative change over time enables the determination of wear in the radiographs.

The measured minimum inlay height of 335 unloaded inlays excluding the weight-induced deformation, served as ground truth for validation and was compared to the algorithmically calculated component distances from 20 radiographs.

With an average weight of 94 kg in the studied TKA patient cohort, it was determined that an average inlay height of 6.160 mm is expected in the patient. Based on the radiographs, the algorithm calculated a minimum component distance of 6.158 mm (SD = 81 µm), which deviated by 2 µm in comparison to the expected inlay height.

An automated method was presented that allows accurate and precise determination of the inlay height and subsequently the wear in knee arthroplasty based on a clinical radiograph and the CAD models. Precision and accuracy are comparable to the current gold standard RSA [2], but without relying on special radiographic setups. The developed method can therefore be used to objectively investigate novel implant materials with meaningful clinical cohorts, thus improving the quality of patient care.

Arthroplasties are widely performed to improve mobility and quality of life for symptomatic knee/hip osteoarthritis patients. With increasing rates of Total Joint Replacements in the United Kingdom, predicting length of stay is vital for hospitals to control costs, manage resources, and prevent postoperative complications. A longer Length of stay has been shown to negatively affect the quality of care, outcomes and patient satisfaction. Thus, predicting LOS enables us to make full use of medical resources.

Clinical characteristics were retrospectively collected from 1,303 patients who received TKA and THR. A total of 21 variables were included, to develop predictive models for LOS by multiple machine learning (ML) algorithms, including Random Forest Classifier (RFC), K-Nearest Neighbour (KNN), Extreme Gradient Boost (XgBoost), and Na¯ve Bayes (NB). These models were evaluated by the receiver operating characteristic (ROC) curve for predictive performance. A feature selection approach was used to identify optimal predictive factors. Based on the ROC of Training result, XgBoost algorithm was selected to be applied to the Test set.

The areas under the ROC curve (AUCs) of the 4 models ranged from 0.730 to 0.966, where higher AUC values generally indicate better predictive performance. All the ML-based models performed better than conventional statistical methods in ROC curves. The XgBoost algorithm with 21 variables was identified as the best predictive model. The feature selection indicated the top six predictors: Age, Operation Duration, Primary Procedure, BMI, creatinine and Month of Surgery.

By analysing clinical characteristics, it is feasible to develop ML-based models for the preoperative prediction of LOS for patients who received TKA and THR, and the XgBoost algorithm performed the best, in terms of accuracy of predictive performance. As this model was originally crafted at Ashford and St. Peters Hospital, we have naturally named it as THE ASHFORD OUTCOME.

Background

Dislocation is a common complication following total hip arthroplasty (THA), and accounts for a high percentage of subsequent revisions. The purpose of this study was to develop a convolutional neural network (CNN) model to identify patients at high risk for dislocation based on postoperative anteroposterior (AP) pelvis radiographs.

The pathogenesis of scoliosis progression remains poorly understood. Seventy-two subject data sets, consisting of four successive values of Cobb-angle and lateral deviations at apices for six and twelve-months intervals in the coronal plane, were used to train and test an artificial neural network (ANN) to predict spinal deformity progression. The accuracies of the trained ANN (3-4-1) for training and testing data were within 3.64° (±2.58°) and 4.40° (±1.86°) of Cobb angles, and within 3.59 (±3.96) mm and 3.98 (±3.41) mm of lateral deviations, respectively. The adapted technique for predicting the scoliosis deformity progression has promising clinical applications.

Scoliosis is a common and poorly understood three-dimensional spinal deformity. The study purpose is to predict scoliosis progression at six and twelve months intervals in the future using successive spinal indices with an artificial neural network (ANN).

The adapted ANN technique enables earlier detection of scoliosis progression with high accuracy. Improved prediction of scoliosis progression will impact bracing or surgical treatment decisions, and may decrease hazardous X-ray exposure.

Seventy-two data sets from adolescent idiopathic scoliosis subjects recruited at the Alberta Children’s Hospital were used in this study. Data sets composed of four successive values of Cobb angles and lateral deviations at apices for six and twelvemonth intervals (coronal plane) were extracted to train and test a specific ANN for predicting scoliosis progression.

Progression patterns in Cobb angles (n = 10) and lateral deviations (n = 8) were successfully identified. The accuracies of the trained ANN (3-4-1) with the training and testing data sets were 3.64° (±2.58°) and 4.40° (±1.86°) of Cobb angles, 3.59 (±3.96) mm and 3.98 (±3.41) mm of lateral deviations, respectively. These results are in close agreement with those using cubic spline extrapolation techniques (3.49° ± 1.85° and 3.31 ± 4.22 mm) and adaptive neuro-fuzzy inference system (3.92° ±3.53° and 3.37 ±3.95 mm) for the same testing data.

ANN can be a promising technique for prediction of scoliosis progression with substantial improvements in accuracy over current techniques, leading to potentially important implications for scoliosis monitoring and treatment decisions.

Funding: AHFMR, CIHR, Fraternal Order of Eagles, NSERC, GEOIDE.

Aims. Precise implant positioning, tailored to individual spinopelvic biomechanics and phenotype, is paramount for stability in total hip arthroplasty (THA). Despite a few studies on instability prediction, there is a notable gap in research utilizing artificial intelligence (AI). The objective of our pilot study was to evaluate the feasibility of developing an AI algorithm tailored to individual spinopelvic mechanics and patient phenotype for predicting impingement. Methods. This international, multicentre prospective cohort study across two centres encompassed 157 adults undergoing primary robotic arm-assisted THA. Impingement during specific flexion and extension stances was identified using the virtual range of motion (ROM) tool of the robotic software. The primary AI model, the Light Gradient-Boosting Machine (LGBM), used tabular data to predict impingement presence, direction (flexion or extension), and type. A secondary model integrating tabular data with plain anteroposterior pelvis radiographs was evaluated to assess for any potential enhancement in prediction accuracy. Results. We identified nine predictors from an analysis of baseline spinopelvic characteristics and surgical planning parameters. Using fivefold cross-validation, the LGBM achieved 70.2% impingement prediction accuracy. With impingement data, the LGBM estimated direction with 85% accuracy, while the support vector machine (SVM) determined impingement type with 72.9% accuracy. After integrating imaging data with a multilayer perceptron (tabular) and a convolutional neural network (radiograph), the LGBM’s prediction was 68.1%. Both combined and LGBM-only had similar impingement direction prediction rates (around 84.5%). Conclusion. This study is a pioneering effort in leveraging AI for impingement prediction in THA, utilizing a comprehensive, real-world clinical dataset. Our machine-learning algorithm demonstrated promising accuracy in predicting impingement, its type, and direction. While the addition of imaging data to our deep-learning algorithm did not boost accuracy, the potential for refined annotations, such as landmark markings, offers avenues for future enhancement. Prior to clinical integration, external validation and larger-scale testing of this algorithm are essential. Cite this article: Bone Jt Open 2024;5(8):671–680

Aims. Hip dysplasia (HD) leads to premature osteoarthritis. Timely detection and correction of HD has been shown to improve pain, functional status, and hip longevity. Several time-consuming radiological measurements are currently used to confirm HD. An artificial intelligence (AI) software named HIPPO automatically locates anatomical landmarks on anteroposterior pelvis radiographs and performs the needed measurements. The primary aim of this study was to assess the reliability of this tool as compared to multi-reader evaluation in clinically proven cases of adult HD. The secondary aims were to assess the time savings achieved and evaluate inter-reader assessment. Methods. A consecutive preoperative sample of 130 HD patients (256 hips) was used. This cohort included 82.3% females (n = 107) and 17.7% males (n = 23) with median patient age of 28.6 years (interquartile range (IQR) 22.5 to 37.2). Three trained readers’ measurements were compared to AI outputs of lateral centre-edge angle (LCEA), caput-collum-diaphyseal (CCD) angle, pelvic obliquity, Tönnis angle, Sharp’s angle, and femoral head coverage. Intraclass correlation coefficients (ICC) and Bland-Altman analyses were obtained. Results. Among 256 hips with AI outputs, all six hip AI measurements were successfully obtained. The AI-reader correlations were generally good (ICC 0.60 to 0.74) to excellent (ICC > 0.75). There was lower agreement for CCD angle measurement. Most widely used measurements for HD diagnosis (LCEA and Tönnis angle) demonstrated good to excellent inter-method reliability (ICC 0.71 to 0.86 and 0.82 to 0.90, respectively). The median reading time for the three readers and AI was 212 (IQR 197 to 230), 131 (IQR 126 to 147), 734 (IQR 690 to 786), and 41 (IQR 38 to 44) seconds, respectively. Conclusion. This study showed that AI-based software demonstrated reliable radiological assessment of patients with HD with significant interpretation-related time savings. Cite this article: Bone Jt Open 2022;3(11):877–884

Aims. Accurate identification of the ankle joint centre is critical for estimating tibial coronal alignment in total knee arthroplasty (TKA). The purpose of the current study was to leverage artificial intelligence (AI) to determine the accuracy and effect of using different radiological anatomical landmarks to quantify mechanical alignment in relation to a traditionally defined radiological ankle centre. Methods. Patients with full-limb radiographs from the Osteoarthritis Initiative were included. A sub-cohort of 250 radiographs were annotated for landmarks relevant to knee alignment and used to train a deep learning (U-Net) workflow for angle calculation on the entire database. The radiological ankle centre was defined as the midpoint of the superior talus edge/tibial plafond. Knee alignment (hip-knee-ankle angle) was compared against 1) midpoint of the most prominent malleoli points, 2) midpoint of the soft-tissue overlying malleoli, and 3) midpoint of the soft-tissue sulcus above the malleoli. Results. A total of 932 bilateral full-limb radiographs (1,864 knees) were measured at a rate of 20.63 seconds/image. The knee alignment using the radiological ankle centre was accurate against ground truth radiologist measurements (inter-class correlation coefficient (ICC) = 0.99 (0.98 to 0.99)). Compared to the radiological ankle centre, the mean midpoint of the malleoli was 2.3 mm (SD 1.3) lateral and 5.2 mm (SD 2.4) distal, shifting alignment by 0.34. o. (SD 2.4. o. ) valgus, whereas the midpoint of the soft-tissue sulcus was 4.69 mm (SD 3.55) lateral and 32.4 mm (SD 12.4) proximal, shifting alignment by 0.65. o. (SD 0.55. o. ) valgus. On the intermalleolar line, measuring a point at 46% (SD 2%) of the intermalleolar width from the medial malleoli (2.38 mm medial adjustment from midpoint) resulted in knee alignment identical to using the radiological ankle centre. Conclusion. The current study leveraged AI to create a consistent and objective model that can estimate patient-specific adjustments necessary for optimal landmark usage in extramedullary and computer-guided navigation for tibial coronal alignment to match radiological planning. Cite this article: Bone Jt Open 2022;3(10):767–776