A predicting model for outcomes of subthalamic nucleus deep brain stimulation in Parkinson’s disease

Objective To establish a predicting model for the therapeutic outcomes of subthalamic nucleus deep brain stimulation (STN-DBS) in Parkinson’s disease (PD) and explore the underlying effect of predictors on DBS outcomes. Methods A retrospective review of clinical data of 31 patients with PD who underwent STN-DBS in our hospitals from January 2013 to December 2013 was performed. Each follow-up as an observation value was integrated into the model. Correlation matrix was used to explore the relationship between each variable and clinical efficacy. Stepwise regression was used to establish a linear regression equation. The normality, variance, linearity and strong influence of the model were evaluated. Results Disease duration, age, preoperative on-state duration, preoperative non-dyskinesia duration, mini-mental state examination scores, following up times, trajectory length through STN and total electric energy delivered were included in the prediction model, in which R2 was 0.433 and F value was 9.345 (P=0.000). The prediction model satisfied the linearity, homoscedasticity, outlier check and influential observation check. Conclusions Our model could effectively predict the therapeutic outcomes of STN-DBS in PD patients. This predicting model could serve as a reference for doctors and the future automatic post-DBS programming. Key words: Parkinson’s disease; Deep brain stimulation; Nubthalamic nucleus; Predicting model