Construction and validation of a nomogram for predicting the risk of severe diabetic oculomotor nerve palsy

Diabetic oculomotor nerve palsy (DONP) mainly manifests as ptosis and diplopia, but limited literature quantitatively assesses its severity. This study analyzed risk factors for severe DONP occurrence, constructed a predictive model, and validated its efficacy. Retrospective analysis was conducted on 180 DONP patients admitted between January 2020 and September 2025. According to the Ocular Motor Nerve Palsy Scale (OMNPS), participants were divided into severe paralysis group (total score ≥12, n = 109) and mild paralysis group (n = 71). The Least Absolute Shrinkage and Selection Operator (LASSO) regression was used for variable screening and multivariate logistic regression for model construction. Bootstrap resampling was used for internal validation. Receiver operating characteristic (ROC) curve, calibration curve and decision curve analysis (DCA) were used to assess model performance. Subgroup analysis was performed to explore clinical characteristics of severe paralysis patients. Age and glycated hemoglobin (HbA1c) were independent risk factors for severe paralysis, while posterior cerebral artery stenosis or plaque served as an auxiliary predictive indicator. The AUC of model was 0.721 (95%CI: 0.645-0.798), with favorable predictive validity confirmed by calibration curve and DCA. In severe cases, each OMNPS score was significantly higher than in mild cases, and scores of ptosis, upward movement, pupil diameter and light reflex strongly correlated with the total score. A nomogram with good discrimination and calibration was constructed to predict severe DONP occurrence. In severe patients, involvement of ptosis, upward motion, and pupillary function exhibited the strongest correlations with disease severity.