Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 7 appraisals
Clinical spine surgery
An Introduction to Machine Learning for the Practicing Spine Surgeon
The applications of new and emerging technologies in spine surgery are constantly expanding. Specifically, machine learning algorithms have seen a rise in utilization in clinical research, allowing for interpretation of large datasets that have the capability of experiential learning. The goal of this work is to present a guide for surgeons to better understand model design, key takeaways, and common pitfalls related to machine learning to ensure accurate and appropriate interpretation of analytical findings in their practice.
2 Aug 2026
Read appraisal →Neurosurgical review
The impact of augmented reality on radiation exposure during spine surgery: a systematic review
BACKGROUND: Augmented reality (AR) has rapidly emerged as an important navigational technology in spine surgery, with growing interest in its potential to reduce dependence on fluoroscopy and mitigate radiation exposure for both patients and operating room staff. METHODS: This systematic review examined current evidence on radiation-related outcomes of AR-assisted spine procedures. A PRISMA-guided search of PubMed, Embase, and Scopus identified twelve eligible studies, including randomized trials, prospective and retrospective clinical cohorts, and cadaveric or synthetic-model investigations. RESULTS: Across the included studies, AR was consistently associated with reduced fluoroscopy use, with several studies demonstrating statistically significant reductions in exposure time compared with conventional fluoroscopy. Two clinical studies directly measured occupational radiation and reported substantially lower staff doses when AR-based navigation and optimized shielding strategies were employed. Patient radiation exposure was similarly decreased in most studies, particularly when AR was integrated with low-dose cone-beam CT protocols. Operative time findings were mixed, reflecting early learning curves and variability in AR systems, but accuracy remained high across platforms. CONCLUSION: Current evidence suggests that AR may reduce intraoperative radiation exposure without compromising workflow or surgical precision, available studies are limited by small sample sizes, heterogenous methodologies, and a paucity of direct staff-dosimetry data. Larger, high-quality multicenter studies are needed to clarify the magnitude of AR's radiation-sparing benefits and to define its role in modern spine surgery.
19 May 2026
Read appraisal →Spine
Identifying Predictors of Failed Back Surgery Syndrome Following Lumbar Spine Surgery: A Machine Learning Approach
STUDY DESIGN: Retrospective cohort study from a tertiary academic medical center. OBJECTIVE: To build a prognostic machine learning model to predict 1-year FBSS incidence after lumbar spine surgery. SUMMARY OF BACKGROUND DATA: A minority of patients who undergo degenerative lumbar spine surgery will have persistent postoperative pain, characterized as "Failed Back Surgery Syndrome" (FBSS). Adequate preoperative identification of patients at risk of having an undesirable outcome after surgery is an essential part of a spine surgeon's workflow. Although several studies have proposed mechanisms and risk factors for FBSS, no studies have developed a prognostic machine learning model to quantify and functionalize predictions. METHODS: A cohort of lumbar fusion and lumbar decompression surgeries was queried from a tertiary academic medical center from 2002 to 2022. Patient and operative characteristics were systematically extracted for each surgery. Several machine learning algorithms were used and optimized to predict FBSS occurrence within 1 year of surgery. SHAP feature importance values were computed for the top-performing model. RESULTS: A total of 10,128 unique lumbar decompression surgeries and 2890 unique lumbar fusion surgeries were included. The Random Forest model had the highest performance of tested models (AUROC of 0.715 for lumbar decompression, 0.701 for lumbar fusion). For lumbar decompression, the top three predictors of FBSS were absence of microdiscectomy, lack of preoperative immunosuppressant usage, and preoperative benzodiazepine usage. For lumbar fusion, prior FBSS diagnosis, lack of preoperative immunosuppressant usage, and operating room duration were the most important predictors. Other key variables spanned several domains, including preoperative medication usage, patient demographics, and operative indications and characteristics. CONCLUSION: This study demonstrates the successful creation of a prognostic machine learning model for prediction of FBSS within one year postoperatively. These models, after external validation, have the potential to be instrumental aspects of a spine surgeon's workflow. LEVEL OF EVIDENCE: Level III.
16 May 2026
Read appraisal →BMJ open
Prediction of Oswestry Disability Index and Numeric Rating Scale scores after lumbar spine surgery: machine learning model development and fairness assessment
BACKGROUND: One-third of patients operated for degenerative conditions in the lumbar spine do not report substantial improvement after 12 months. Most previous outcome prediction models are classifiers. This constrains nuances in prediction and use for decision support. OBJECTIVES: To develop and test models for the prediction of continuous outcome scores and retrieval of similar patients' outcomes, and to evaluate the models' fairness. SETTING: Norwegian public and private specialist healthcare. PARTICIPANTS AND DATA SOURCE: All cases recorded with an elective operation for lumbar disc herniation (LDH, n=18 377) or lumbar spinal stenosis (LSS, n=24 540) in the Norwegian Registry for Spine Surgery from 1 January 2007 to 23 May 2023. OUTCOME MEASURES: All outcomes were patient-reported 12 months after the operation. The primary outcome was the Oswestry disability index (ODI), modelled on a scale ranging from 0 to 100. Numeric Rating Scale scores (range 0-10) for back and leg pain were secondary outcomes. MODEL BUILDING AND PERFORMANCE: We selected 22 predictors recorded preoperatively by patients and clinicians based on Shapley Additive Explanations values. Data were split into 80%/20% training/test samples for LDH and LSS. Six machine learning methods for regression, that is, with a continuous outcome (extreme gradient boosting (XGBoost), Gaussian process regression, gradient boosting regression, artificial neural networks and linear regression), were trained for both conditions using fivefold cross-validation. We report the magnitude and distribution of errors as mean absolute error (MAE) with 95% CIs, and explanatory power as the coefficient of determination (R2). Fairness and calibration were assessed with violin and calibration plots of error. We developed a patient-similarity function that uses a K-nearest neighbour model to retrieve the individual outcomes of the 50 most similar patients and evaluated it by calculating L1 distances (Manhattan distances) across subgroups. RESULTS: XGBoost regression performed best for both conditions. The models showed good calibration and predicted ODI with MAE 11.32 (95% CI 11.00 to 11.63) and R2 0.27 (95% CI 0.24 to 0.29) for LDH and MAE 12.05 (95% CI 11.76 to 12.32) and R2 0.31 (95% CI 0.28 to 0.34) for LSS. The MAEs for back and leg pain were 2.09 (95% CI 2.04 to 2.15) and 1.95 (95% CI 1.90 to 2.00) for LDH and 2.33 (95% CI 2.28 to 2.38) and 2.13 (95% CI 2.08 to 2.16) for LSS. All models were fair with differences in error between subgroups for sex, age, education level and native language. In the patient-similarity function, distances at baseline were evenly distributed across subgroups. CONCLUSIONS: Our machine learning models predicted continuous outcomes with MAEs close to the SEs of measurements. The models were fair across sociodemographic subgroups. We succeeded in developing a patient-similarity function which supplements the predictions.
15 May 2026
Read appraisal →AJNR. American journal of neuroradiology
High-Resolution 2D versus 3D Lumbar Spine MRI Optimized with a Deep Learning Reconstruction Algorithm and Prototype Conformal Coil
BACKGROUND AND PURPOSE: Lumbar spine MRI is predominantly performed using 2D FSE sequences. 3D FSE sequences offer potential advantages over 2D, especially with recent advances in deep-learning reconstruction (DLR) and the use of conformal high-density coil arrays to improve the SNR. This study aimed to compare image quality and diagnostically relevant differences between 2D and 3D lumbar spine protocols, both using a prototype conformal coil and DLR. MATERIALS AND METHODS: Subjects with lumbar curvature referred for routine lumbar spine MRI were prospectively recruited (n=31). A prototype, conformal spine coil array with 70 total posterior receiver elements was used at 3T. DLR was applied to both the 2D- and 3D-protocols to improve sharpness and reduce noise. For 3D, a prototype DLR ("Sonic DL") technique was applied to provide higher acceleration factors. Three attending musculoskeletal radiologists assessed 2D and 3D T1-weighted (T1), T2-weighted (T2), and STIR sequences for image quality and diagnostic performance. The Gwet agreement coefficient (AC1, AC2) or intraclass correlation (ICC) was used to assess interrater agreement, while the Wilcoxon signed rank test or a sign test with Bonferroni correction assessed differences between 2D and 3D sequences. RESULTS: There was an overall 14% scan time reduction of 3D compared with 2D sequences, and a 38% time reduction for T2 sequences. All 3D sequences exhibited similar image quality to 2D. 3D T2 sequences had less CSF flow artifact than 2D T2 sequences. Interrater agreement was mostly comparable between all 2D and 3D assessments, including excellent agreement for the spinal canal area (ICC = 0.952 [95% CI: 0.907-0.976] for 2D, 0.941 [0.866-0.973] for 3D), good agreement for foraminal stenosis assessments on T2 sequences (AC2 = 0.713 to 0.795 for 2D, 0.706-0.832 for 3D), and good agreement for fracture detection on T1 sequences (AC1 = 0.869 for 2D, 0.815 for 3D). CONCLUSIONS: DLR-enabled 3D lumbar spine MRI offers comparable image quality to 2D with reduced overall scan time.
11 May 2026
Read appraisal →Neurosurgical review
Sequesterectomy versus standard micro‑/open discectomy for lumbar disc herniation: a systematic review and meta-analysis
Lumbar disk herniation is a common clinical entity affecting about 1% of the entire population every year. The purpose of the present work is to provide a systematic review and meta-analysis comparing the safety and efficiency of sequesterectomy and discectomy in the management of lumbar disk herniation. A systematic review and meta-analysis using the random effects method was performed. This search was applied through the 6th of March 2025 to PubMed, Scopus, Cochrane Central Register of Controlled Trials and the Directory of Open Access Journals. The data collected included patient demographics, spinal level at which the operation occurred, duration of operation and hospital stay, outcome, recurrence rate as well as potential immediate and delayed complications. A total of 18 original studies with a cumulative number of 4394 patients were identified. No statistically significant difference was found in the re-herniation and complication rate between the two groups (OR: 1.058, 95% CI: 0.726-1.541, p=.769 and OR: 1.399, 95% CI: 0.964-2.032, p=.077 respectively). The same applied for mean hospital stay (standard difference in means: 0.206, SE = 0.167, p= .219). In contrast, mean operating time was found to be significantly shorter in the sequesterectomy group than in the discectomy group (mean 58.086 min versus 72.885 min respectively). As or the patient reported outcomes, VAS score for back pain two years after the operation was the only index that was significantly higher in patients with discectomy (standard difference in means: 0.348, SE = 0.172, p= .04). Based on the available data, the two approaches do not differ in terms of risk of reherniation, reoperation rate and postoperative complications. Nevertheless, patients that have undergone sequesterectomy could potentially benefit from a more significant reduction in back pain in the long run.
10 May 2026
Read appraisal →PloS one
Controlled comparative study of YOLOv8-Pose, YOLOv11-Pose, and Detectron2 for vertebrae detection and keypoint estimation
Accurate vertebrae detection with precise keypoint localization is essential for medical image analysis and clinical applications, such as spinal alignment assessment. This study presents a controlled, task-driven comparison of four pose-based deep learning models for vertebrae detection and keypoint estimation: YOLOv8n-Pose, YOLOv11n-Pose, and Detectron2 (Keypoint R-CNN) with ResNet-50 and ResNet-101 backbones. A single-class vertebrae dataset with bounding box annotations and four anatomically meaningful keypoints per vertebra was evaluated under consistent training settings. Performance was assessed in terms of keypoint localization accuracy, detection precision, inference speed, and vertebrae-specific prediction behavior, including detection completeness and duplicate detections, which are critical for clinical usability. Results show that YOLOv11n-Pose provides the best balance between keypoint accuracy and inference efficiency, while YOLOv8n-Pose achieves the fastest inference with competitive performance. Detectron2-based models exhibit lower keypoint accuracy, slower inference, and frequent duplicate predictions that obscure correct anatomical landmarks. Additional experiments indicate that larger YOLO variants (particularly YOLOv8l-Pose) improve accuracy, albeit at increased computational cost. Overall, the results demonstrate that model capacity or architectural recency alone does not guarantee superior performance in vertebrae keypoint detection, highlighting the importance of anatomy-aware model selection for spine imaging applications.
23 Apr 2026
Read appraisal →