Personalized short-term prognostic nomogram model for patients with Parkinson's disease treated with subthalamic nucleus deep brain stimulation

ZHU Qiushi; LIU Huaixin; GENG Fengyang; LIU Peihui; JIANG Yang

doi:10.3969/j.issn.2097-1753.2023.06.008

Abstract：Objective To construct an individualized short-term prognostic nomogram model for Parkinson's disease(PD)patients undergoing subthalamic nucleus-deep brain stimulation(STN-DBS),and to assess the clinical accuracy and applicability of this model.Methods Parkinson's disease patients who underwent STN-DBS treatment at Brain Hospital of Liaocheng People's Hospital between January 1,2015 and January 1,2022 were randomly selected and followed up for one year except those were out of contact.Patients were divided into the good prognosis group(S&E score＞70 points)and the poor prognosis group(S&E score ≤70 points)based on the Schwab and England Scale(S&E score)related to the medication phase.Independent risk factors for poor short-term prognosis of PD patients undergoing deep brain stimulation were analyzed using multifactorial logistics analysis.A nomogram model was constructed using R language based on the above independent riskfactors.The Bootstrap method was employed to validate the accuracy of prediction of the model,while the receiver operating characteristic(ROC)curve was used to evaluate the predictability of the model.The decision curve analysis was used to assess the clinical benefit of the model.Results There were 50 patients in the good prognosis group and 35 in the poor prognosis group.Identified through multifactorial logistics analysis,independent risk factors leading to poor short-term prognosis of STN-DBS treatment for PD patients included the age of onset,Hoehn-Yahr staging,surgical age,mini-mental state examination(MMSE)score,and Unified Parkinson's Disease Rating Scale Ⅱ(UPDRS Ⅱ)score.The prognostic nomogram model constructed based on these independent risk factors demonstrated strong predictability,with an area under the curve(AUC)of 0.88,95％CI:0.71-0.94,according to the ROC curve.Bootstrap validation of the model-related data showed good accuracy of prediction:C-index 0.867,95％CI:0.814-0.928.The calibration plot predicted an average absolute error of 0.011 for the risk of poor short-term prognosis of STN-DBS treatment for PD patients.The Hosmer-Leme show test results were statistically insignificant(x2=5.062,P=0.663).The clinical benefit curve showed optimal clinical benefit and applicability of the model when the threshold for MACCE occurrence was between 0.11 and 0.97.Conclusion The age of onset,Hoehn-Yahr staging,surgical age,MMSE,and UPDRSⅡ are independent risk factors for poor short-term prognosis of STN-DBS treatment for PD patients.The nomogram model constructed here is highly accurate and applicable.