Abstract
Somatosensory technology offers an innovative approach for motor function rehabilitation in children with low vision. However, limited studies have explored the key factors influencing its effectiveness, and predictive models for clinical guidance are lacking. This study included 183 children with low vision who underwent somatosensory-based motor rehabilitation training from January 2021 to December 2023. Participants were randomly assigned to a training set (n = 128) and a validation set (n = 55). Multivariate logistic regression was used to identify independent factors associated with rehabilitation outcomes. A nomogram prediction model was developed based on significant variables. Model performance was assessed using ROC curves, calibration plots, the Hosmer-Lemeshow test, and decision curve analysis (DCA). Rehabilitation was ineffective in 38 children (29.69%) in the training set and 15 children (27.27%) in the validation set. Independent risk factors for poor outcomes included long disease duration, low family support, presence of complications, low training intensity, poor baseline vision, and limited adaptability to somatosensory technology (P < 0.05). The nomogram showed excellent predictive performance, with C-index values of 0.935 and 0.841, and AUCs of 0.941 and 0.843 in the training and validation sets, respectively. Calibration curves indicated good agreement between predicted and observed outcomes. DCA demonstrated clinical utility across a wide range of threshold probabilities. The nomogram effectively predicts rehabilitation outcomes in children with low vision receiving somatosensory-based training. This tool can assist clinicians in developing individualized rehabilitation strategies and improving intervention effectiveness.