JHS - 2026-10-19 - Journal Article
An Explainable Prediction Model for Outcomes After Anterior Subcutaneous Ulnar Nerve Transposition in Idiopathic Cubital Tunnel Syndrome.
Lim JH, Kim MS
Topics
Key Takeaway
Wrist-level CMAP amplitude ≥6.6 mV and modified McGowan grade ≤IIA are the dominant predictors of excellent/good outcome after anterior subcutaneous ulnar nerve transposition, with only 57.1% of 119 patients achieving that threshold.
Summary Depth
Choose how much analysis to show on this article page.
Summary
This retrospective cohort evaluated preoperative predictors of outcome after anterior subcutaneous ulnar nerve transposition for idiopathic cubital tunnel syndrome in 119 patients using conventional regression and six machine-learning algorithms. The extremely randomized trees model achieved the highest AUC and lowest Brier score among all algorithms tested. Wrist-level CMAP amplitude ≥6.6 mV and modified McGowan grade ≤IIA emerged as the dominant prognostic features for excellent/good outcome by Messina's criteria.
Key Limitation
Outcomes were dichotomized using Messina's criteria without reporting continuous functional scores, potentially obscuring clinically meaningful gradations within the fair/poor group and limiting comparison to studies using DASH, QuickDASH, or MCID-based endpoints.
Original Abstract
PURPOSE
To identify preoperative clinical and electrodiagnostic predictors of outcome after anterior subcutaneous ulnar nerve transposition for idiopathic cubital tunnel syndrome and to develop an explainable model for individualized prognosis.
METHODS
This retrospective cohort study included 119 patients who underwent anterior subcutaneous ulnar nerve transposition for idiopathic cubital tunnel syndrome between 2005 and 2022 with at least 1 year of follow-up. Candidate predictors included demographic factors, occupational category score, elbow carrying angle, modified McGowan grade, and electrodiagnostic parameters. Outcomes were graded using Messina's criteria and dichotomized as excellent/good versus fair/poor for prediction modeling. Conventional regression and six machine-learning algorithms were evaluated, with the machine-learning models internally validated using nested stratified 5-fold cross-validation repeated 50 times. Model performance was assessed using the area under the receiver operating characteristic curve, F1 score, and Brier score. Interpretation relied on feature importance and local interpretable model-agnostic explanations.
RESULTS
Sixty-eight patients (57.1%) had an excellent/good outcome, whereas 51 patients (42.9%) had a fair/poor outcome. The extremely randomized trees model achieved the highest mean area under the receiver operating characteristic curve and the lowest Brier score and was selected for model interpretation. Wrist-level compound muscle action potential amplitude and modified McGowan grade were the dominant prognostic features. Exploratory model interpretation supported wrist-level compound muscle action potential amplitude ≥6.6 mV and modified McGowan grade ≤IIA as reference values for an excellent/good outcome.
CONCLUSIONS
Wrist-level compound muscle action potential amplitude and preoperative clinical severity were the key predictors of recovery after anterior subcutaneous transposition for idiopathic cubital tunnel syndrome. Among the evaluated algorithms, the extremely randomized trees model achieved the highest mean area under the receiver operating characteristic curve and the lowest mean Brier score, reinforcing the prognostic importance of these variables. Explainable prediction modeling may support individualized preoperative counseling.
TYPE OF STUDY/LEVEL OF EVIDENCE
Prognostic III.