Tag: Risk prediction

Posted on

Development and Validation of Interpretable Machine Learning Models Incorporating Paraspinal Muscle Quality to Predict Cage Subsidence Risk Following Posterior Lumbar Interbody Fusion

Spine 2025;50:1375–1385

This multicenter retrospective study developed and validated an interpretable LightGBM machine learning model incorporating paraspinal muscle quality and bone metrics to accurately predict cage subsidence risk after PLIF. Key risk factors included lower psoas muscle index, higher fat infiltration, reduced bone density, and suboptimal cage parameters.

• A machine learning model (LightGBM) was developed to predict cage subsidence risk after PLIF, achieving high accuracy (AUC 0.9752, 92% accuracy, F1 score 0.92).

• Key independent risk factors include lower psoas muscle index (PMI), higher fat infiltration (FI), reduced bone density (HU value, VBQ), suboptimal cage position/height, and greater postoperative changes in intervertebral height (IH) and segmental angle (SA).

• Paraspinal muscle quality was a major contributor; removing muscle indicators reduced model accuracy substantially.

• Patients with cage subsidence had poorer paraspinal muscle and bone quality compared to those without subsidence.

• The model was externally validated and deployed as a web-based tool for real-time, individualized clinical risk assessment.

• Findings support personalized surgical planning and risk mitigation strategies for PLIF patients.

• The study emphasizes a multifactorial approach, integrating skeletal, muscular, and surgical parameters for optimal prediction.

• Limitations include retrospective design, use of a single cage type, and lack of comorbidity indices; further prospective studies are needed.

Posted on

Validation of the Supplemented Spetzler-Martin Grading System for Brain Arteriovenous Malformations in a Multicenter Cohort of 1009 Surgical Patients

AVM

Neurosurgery 76:25–33, 2015

The supplementary grading system for brain arteriovenous malformations (AVMs) was introduced in 2010 as a tool for improving preoperative risk prediction and selecting surgical patients.

OBJECTIVE: To demonstrate in this multicenter validation study that supplemented Spetzler-Martin (SM-Supp) grades have greater predictive accuracy than Spetzler-Martin (SM) grades alone.

METHODS: Data collected from 1009 AVM patients who underwent AVM resection were used to compare the predictive powers of SM and SM-Supp grades. Patients included the original 300 University of California, San Francisco patients plus those treated thereafter (n = 117) and an additional 592 patients from 3 other centers.

RESULTS: In the combined cohort, the SM-Supp system performed better than SM system alone: area under the receiver-operating characteristics curve (AUROC) = 0.75 (95% confidence interval, 0.71-0.78) for SM-Supp and AUROC = 0.69 (95% confidence interval, 0.65-0.73) for SM (P , .001). Stratified analysis fitting models within 3 different follow-up groupings (,6 months, 6 months-2 years, and .2 years) demonstrated that the SM-Supp system performed better than SM system for both medium (AUROC = 0.71 vs 0.62; P = .003) and long (AUROC = 0.69 vs 0.58; P = .001) follow-up. Patients with SMSupp grades #6 had acceptably low surgical risks (0%-24%), with a significant increase in risk for grades .6 (39%-63%).

CONCLUSION: This study validates the predictive accuracy of the SM-Supp system in a multicenter cohort. An SM-Supp grade of 6 is a cutoff or boundary for AVM operability. Supplemented grading is currently the best method of estimating neurological outcomes after AVM surgery, and we recommend it as a starting point in the evaluation of AVM operability.

Proudly powered by WordPress