Machine learning, decision tree and nomogram for predicting screw loosening after PLIF in osteoporotic patients: a retrospective multicenter study

An interpretable multimodal biomechanical–radiological model for predicting fixation failure in osteoporotic hip fractures: A retrospective cohort study

This study aimed to develop and validate a multimodal prediction model integrating biomechanical and radiological variables to predict internal fixation failure in osteoporotic hip fractures, enabling individualized risk assessment and perioperative decision-making. Patients with osteoporotic hip fractures undergoing internal fixation between March 2019 and February 2024 were retrospectively enrolled and randomly divided into a training set (n = 249) and a validation set (n = 107) at a 7:3 ratio. The primary outcome was implant-related failure within 12 months post-surgery. In the training set, univariate analysis and multivariate logistic regression were performed to screen associated factors. Using independent predictors, 3 machine learning models (random forest, support vector machine, and K-nearest neighbors) were developed and compared. The model with the best discriminative ability, assessed by the area under the receiver operating characteristic curve (AUC) with internal validation (bootstrapping), calibration curves, and decision curve analysis, was selected to construct a nomogram. No significant differences in baseline characteristics were observed between the training and validation sets (P > .05). Multivariate logistic regression identified that bone mineral density, maximum fracture end displacement, peak stress distribution of the implant, fracture reduction alignment deviation, and implant insertion depth were significantly associated with fixation failure (P < .05). The nomogram demonstrated excellent performance in both the training (AUC = 0.887, 95% confidence interval: 0.835–0.939) and validation sets (AUC = 0.869, 95% confidence interval: 0.801–0.937). Calibration curves showed good agreement between predicted and observed risks (P > .05), and decision curve analysis indicated superior clinical net benefit across a wide threshold range. Stability testing confirmed no significant multicollinearity (variance inflation factor < 2, events per variable ≥ 5). The random forest model demonstrated strong discriminatory ability and calibration in predicting fixation failure in this single-center retrospective cohort. While the model shows promise for perioperative risk stratification, external validation in multicenter prospective studies is required before clinical implementation.

Development of an automated machine learning-based risk prediction and decision support system for postoperative cubitus varus complicating pediatric lateral humeral condyle fracture

Purpose Postoperative cubitus varus is a disabling complication of pediatric lateral humeral condyle fractures that impairs long-term elbow function. This study aimed to develop and validate the first Improved LangEvin Equation-based Evolutionary (ILEE) automated machine learning (AutoML) model and visual decision support system for preoperative prediction of this complication, to enable personalized risk stratification and targeted intervention. Methods We conducted a retrospective cohort study of 330 children with lateral humeral condyle fractures treated between January 2005 and June 2022. An ILEE-optimized AutoML model was constructed and compared with the original LEE algorithm and 6 conventional machine learning models. Model performance was evaluated using area under the receiver operating characteristic curve (ROC-AUC), accuracy, sensitivity, and specificity. Key predictors were identified and interpreted via SHapley Additive exPlanations (SHAP) analysis, and a clinical decision support system was developed using MATLAB App Designer. Results The ILEE algorithm outperformed all comparators in optimization stability and convergence speed. The AutoML model identified five key features for predicting cubitus varus and exhibited better prediction calibration performance compared to other models (ROC-AUC = 0.9557). SHAP analysis ranked the importance of these features as follows: obesity degree, preoperative timing, internal fixation time, fracture type, and external fixation time. The decision support system can generate real-time risk levels, predicted probabilities, and personalized clinical recommendations within 1 s. Conclusion This study establishes the first ILEE-based AutoML model for predicting postoperative cubitus varus in pediatric lateral humeral condyle fractures, demonstrating competitive predictive performance. The user-friendly visual system enables rapid preoperative risk assessment, assisting clinicians in identify high-risk patients, optimize treatment plans, and potentially improving pediatric orthopedic outcomes.

Multivariate Risk Factor Analysis and Clinical Prediction Model Construction of Intercostal Neuralgia Following Single-Segment Osteoporotic Thoracic Vertebral Compression Fractures

Objective To explore the multiple risk factors for intercostal neuralgia after osteoporotic thoracic vertebral compression fractures (OVCF), and to construct a clinical prediction model. Methods The clinical data of 280 patients with single-segment thoracic OVCF admitted to our orthopedic department from January 2022 to December 2025 were retrospectively collected. Patients were categorized into neuropathic pain (NP) and non-neuropathic pain (non-NP) groups based on the Leeds Assessment of Neuropathic Symptoms and Signs (LANSS) scale, with a score of ≥12 defining the primary outcome of intercostal neuralgia as neuropathic pain. Demographic data, fracture-related parameters, bone density, bone metabolism markers, and psychological status (Hospital Anxiety and Depression Scale, HADS) were collected. Univariate and multivariate Logistic regression analyses were used to screen independent risk factors, and the discrimination of the prediction model (area under the ROC curve, AUC) was evaluated. The model’s performance was internally validated using the Bootstrap method. Results Among the 280 patients, the incidence of neuropathic pain was 21.8% (61/280). Multivariate analysis identified six independent risk factors: middle thoracic fracture (T5-T8, OR=4.603), thoracolumbar fascia injury (TLFI, OR=4.883), injured vertebral width ratio (per 0.1 increase, OR=3.973), decreased bone mineral density T-score (per 1-unit decrease, OR=2.685), intravertebral vacuum cleft (IVC, OR=2.764), and depressive state (HADS≥8, OR=2.586). The prediction model showed good calibration (Hosmer-Lemeshow P=0.412) and discrimination (AUC=0.843, 95% CI: 0.789–0.897), with sensitivity 80.3%, specificity 76.7%, and negative predictive value 93.2%. Bootstrap internal validation yielded an optimism-corrected AUC of 0.831. Conclusion Post-fracture intercostal neuralgia in osteoporotic thoracic vertebrae is multifactorial. Middle thoracic vertebra fractures, injury of the thoracolumbar fascia, increased ratio of injured vertebra width, decreased bone density, intravertebral vacuum fissure, and depressive state are independent risk factors. However, external validation in prospective multicenter studies is required before routine clinical implementation.

Machine learning-based prediction of perioperative complications in spine surgery: a large-scale model development and validation study.

Relevant factors of lumbar compression fracture patients’ death 1 year after surgery and its prediction model creation and validation

This study aimed to investigate the risk factors associated with 1-year postoperative mortality in patients with lumbar compression fractures and to construct and validate a predictive nomogram model. Clinical data of patients admitted between January 2021 and December 2024 were retrospectively analyzed. Independent predictors of 1-year mortality were identified using univariate and multivariate logistic regression analyses. A nomogram was constructed based on the final model. Model discrimination was evaluated using the receiver operating characteristic curve and the area under the curve. Calibration, Bootstrap resampling, and 10-fold cross-validation were used for internal validation. A total of 378 patients were included, of whom 21 (5.56%) died within 1 year postoperatively. Five independent predictors were identified: bone mineral density ≤ −2.5, multiple segmental fractures, age > 70 years, albumin ≤ 40 g/L, and neutrophil-to-lymphocyte ratio > 4. The nomogram showed good discriminative performance, with an area under the curve of 0.826 in the training cohort and 0.813 in the validation cohort. Calibration curves demonstrated good agreement between predicted and observed outcomes. One-year postoperative mortality in lumbar compression fracture patients is influenced by multiple clinical and inflammatory factors. The proposed nomogram demonstrates good discriminative ability and may help clinicians identify high-risk patients for early intervention.

Clinical and radiographic factors associated with surgical approach selection in total hip arthroplasty: a preliminary machine learning analysis

Objective To investigate the clinical and radiographic factors associated with surgeon selection of the direct anterior approach (DAA) versus the posterolateral approach (PLA) in total hip arthroplasty (THA), and to explore whether machine learning methods can characterize historical surgical selection patterns. Methods A retrospective analysis was performed on 98 patients who underwent primary THA at two institutions, including 61 patients in the PLA group and 37 patients in the DAA group. Patient demographics and preoperative radiographic parameters were collected. Baseline comparisons and univariate logistic regression analyses were conducted using SPSS. A machine learning classification model reflecting historical surgeon-selected approaches was constructed using the XGBoost algorithm, and its performance was evaluated via receiver operating characteristic (ROC) curves, five-fold cross-validation, SHAP analysis, and calibration curves. Results Patients in the DAA group exhibited significantly higher age, soft tissue thickness, neck-shaft angle (NSA), and femoral offset compared to those in the PLA group (all P < 0.05). Univariate logistic regression revealed that age, NSA, and femoral offset were significantly associated with DAA selection. The XGBoost model achieved an area under the curve (AUC) of 0.938 and an accuracy of 90.0% on the test set, with a mean AUC of 0.897 via five-fold cross-validation. SHAP analysis identified osteoporosis, Dorr classification, age, and NSA as key contributors to model predictions. The model exhibited good calibration, as indicated by a Brier score of 0.0898. Conclusion The present study identified several clinical and radiographic factors associated with historical surgeon selection of DAA versus PLA in THA. The machine learning model demonstrated the ability to characterize real-world surgical selection patterns, although it should not be interpreted as recommending the optimal surgical approach for individual patients.