In recent years, machine learning (ML) and artificial neural networks (ANNs), a particular subset of ML, have been adopted by various areas of healthcare. A number of diagnostic and prognostic algorithms have been designed and implemented across a range of orthopaedic sub-specialties to date, with many positive results. However, the methodology of many of these studies is flawed, and few compare the use of ML with the current approach in clinical practice. Spinal surgery has advanced rapidly over the past three decades, particularly in the areas of implant technology, advanced surgical techniques, biologics, and enhanced recovery protocols. It is therefore regarded an innovative field. Inevitably, spinal surgeons will wish to incorporate ML into their practice should models prove effective in diagnostic or prognostic terms. The purpose of this article is to review published studies that describe the application of neural networks to spinal surgery and which actively compare ANN models to contemporary clinical standards allowing evaluation of their efficacy, accuracy, and relatability. It also explores some of the limitations of the technology, which act to constrain the widespread adoption of neural networks for diagnostic and prognostic use in spinal care. Finally, it describes the necessary considerations should institutions wish to incorporate ANNs into their practices. In doing so, the aim of this review is to provide a practical approach for spinal surgeons to understand the relevant aspects of neural networks.

Cite this article: Bone Joint J 2021;103-B(9):1442–1448.

Aims

The aim of this study was to investigate if there are differences in outcome between sliding hip screws (SHSs) and intramedullary nails (IMNs) with regard to fracture stability.

Aims

The morphology of medial malleolar fracture is highly variable and difficult to characterize without 3D reconstruction. There is also no universally accepeted classification system. Thus, we aimed to characterize fracture patterns of the medial malleolus and propose a classification scheme based on 3D CT reconstruction.

Aims

Pelvic incidence (PI) is considered an important anatomical parameter for determining the sagittal balance of the spine. The contribution of an abnormal PI to hip osteoarthritis (OA) remains controversial. In this study, we aimed to investigate the relationship between PI and hip OA, and the difference in PI between hip OA without anatomical abnormalities (primary OA) and hip OA with developmental dysplasia of the hip (DDH-OA).

Aims

Hip displacement, common in patients with cerebral palsy (CP), causes pain and hinders adequate care. Hip reconstructive surgery (HRS) is performed to treat hip displacement; however, only a few studies have quantitatively assessed femoral head sphericity after HRS. The aim of this study was to quantitatively assess improvement in hip sphericity after HRS in patients with CP.

Aims

Malreduction of the syndesmosis has been reported in up to 52% of patients after fixation of ankle fractures. Multiple radiological parameters are used to define malreduction; there has been limited investigation of the accuracy of these measurements in differentiating malreduction from inherent anatomical asymmetry. The purpose of this study was to identify the prevalence of positive malreduction standards within the syndesmosis of native, uninjured ankles.

Aims

The aim of this study was to determine the differences in spinal imaging characteristics between subjects with or without lumbar developmental spinal stenosis (DSS) in a population-based cohort.

Aims

In the last decade, interest in partial knee arthroplasties and bicruciate retaining total knee arthroplasties has increased. In addition, patient-related outcomes and functional results such as range of movement and ambulation may be more promising with less invasive procedures such as bicompartmental arthroplasty (BCA). The purpose of this study is to evaluate clinical and radiological outcomes after a third-generation patellofemoral arthroplasty (PFA) combined with a medial or lateral unicompartmental knee arthroplasty (UKA) at mid- to long-term follow-up.

We have analysed the Constant-Murley (1987) assessment for 25 patients with shoulder pathology. We found the score easy to use, with low inter- and intraobserver errors, but sufficiently imprecise in repeated measurements to give concern in its use for clinical follow-up of patients. We have calculated 95% confidence limits for a single assessment to be within 16 to 20 points in most cases. In addition, we found that all our subjects with instability as their main problem scored within five points of the maximum; this suggests that the scoring method may need to be revised for use on these patients

Aims

Morphological abnormalities are present in patients with developmental dysplasia of the hip (DDH). We studied and compared the pelvic anatomy and morphology between the affected hemipelvis with the unaffected side in patients with unilateral Crowe type IV DDH using 3D imaging and analysis.

Aims

The aim of this study was to identify the origin and development of the threshold for surgical intervention, highlight the consequences of residual displacement, and justify the importance of accurate measurement.

Accurate measurement of the alignment of the tibia is important both clinically and in research. The conventional method of measuring the angle of malunion after a fracture of the shaft of the tibia is potentially inaccurate because the mechanical axis of the normal bone may not pass down the centre of the medullary canal. An alternative method is described in which a radiograph of the opposite tibia is used as a template. A sample of 56 sets of standard radiographs of healed fractures of the shaft of the tibia was evaluated. The 95% limits of agreement between this and the conventional method were wide, being −6.2° to +5.5° for coronal angulation and −6.7° to +8.1° for sagittal angulation. These results suggest that the conventional method is inaccurate. The new method has good inter- and intraobserver reliability

We studied the reliability of the Singh classification of trabecular bone structure in the proximal femur as a measure of osteoporosis, using kappa statistics. Radiographs of fractures of the femoral neck or trochanteric region in 80 consecutive patients were assessed by six observers. The interobserver variation was large; only three of 72 radiographs were given the same classification by all six observers and the kappa values ranged from 0.15 to 0.54. The intraobserver variation showed substantial strength of agreement; kappa values ranged from 0.63 to 0.88. In 77 patients dual-energy X-ray absorptiometry was used to measure bone mineral density. The results were compared with those of the Singh classification: we found no correlation

The variability in measurement of angles in congenital scoliosis is not known, but it is postulated that it is larger than that in adolescent idiopathic scoliosis due to skeletal immaturity, incomplete ossification, and anomalous development of the end-vertebrae. To determine this variability, we selected 54 radiographs of adequate quality showing 67 scoliotic curves from children with congenital scoliosis. The end-vertebrae were preselected. Each curve was measured by the Cobb method on two separate occasions by six different observers, using the same goniometer and marker. The intraobserver variability was +/- 9.6 degrees and the interobserver variability +/- 11.8 degrees. If 'significant progression' is to be used as a criterion for surgical fusion in congenital scoliosis, there should be at least a 23 degrees increase, the entire range of the interobserver variability, in the curvature to ensure that the perceived increase is not due to variability in measurement

Aims

Rotational acetabular osteotomy (RAO) has been reported to be effective in improving symptoms and preventing osteoarthritis (OA) progression in patients with mild to severe develomental dysplasia of the hip (DDH). However, some patients develop secondary OA even when the preoperative joint space is normal; determining who will progress to OA is difficult. We evaluated whether the preoperative cartilage condition may predict OA progression following surgery using T2 mapping MRI.

We analysed 42 weight-bearing lateral radiographs of the ankle, 20 of which were from patients with a clinical and plain radiological diagnosis of talocalcaneal coalition (TCC) who subsequently had CT. The remainder were from 22 healthy volunteers with no clinical findings suggestive of hindfoot pathology. Four observers, blinded to the CT findings, independently evaluated the radiographs on two separate occasions. With the 95% confidence interval and using the CT findings as the comparison we calculated the sensitivity, specificity, accuracy, and positive and negative predictive values for the C-sign, and for other signs known to be associated with TCC. Similarly, we also calculated the interobserver and intraobserver reliability for these signs using the kappa statistic. Our results suggest that the C-sign is highly sensitive and specific for TCC. It is an accurate indicator and significantly more reliable than other previously recognised radiological signs of TCC. Features of the C-sign, however, cannot be relied upon to indicate whether the TCC is fibrous or bony

We have studied the intervertebral discs adjacent to fractured vertebral bodies using MRI in 63 patients at a minimum of 18 months after injury. There were 75 thoracolumbar fractures of which 26 were treated conservatively and 37 by posterior reduction and fusion with an AO internal fixator. We identified six different types of disc using criteria based on the morphology and the intensity of the MRI signal. The inter- and intraobserver variability of this system was good. Most of the discs showed predominantly morphological changes with no variation in signal intensity. Some disc types were associated with progressive kyphosis in patients treated conservatively. In those managed by operation, recurrent kyphosis appeared to result from creeping of the disc in the central depression of the bony endplate rather than from disc degeneration. Changes in the disc space after posterior fixation should not be seen as a form of chronic instability but as a redistribution of the disc tissue in the changed morphology of the space after fractures of the endplate

Aims

It has been hypothesized that proximal radial neck resorption (PRNR) following press-fit radial head arthroplasty (RHA) is due to stress-shielding. We compared two different press-fit stems by means of radiographs to investigate whether the shape and size of the stems are correlated with the degree of PRNR.

Aims

To study the associations of lumbar developmental spinal stenosis (DSS) with low back pain (LBP), radicular leg pain, and disability.