Construction of an auxiliary identification model for Parkinson's disease based on clinical blood bio-markers

HAN Jiaqi; AN Yu; SUN Mengge

doi:10.3969/j.issn.2096-9058.2025.01.001

Abstract：Objective To explore blood biomarkers related to Parkinson's disease(PD)in clinical laboratories,to find potential PD blood biomarkers and to construct a PD auxiliary prediction model.Methods A total of 216 PD pa-tients treated in the department of Neurology and Functional Neurology of Shanghai East Hospital from January 2019 to Oc-tober 2023 were retrospectively collected as the PD cohort,and 216 healthy subjects matched by gender and age during the same period were randomly selected as the control cohort.Demographic characteristics and blood biomarker results of the PD and control cohorts were recorded.Based on the two data sets,LASSO regression was used to screen PD characteristic biomarkers.The two data sets were randomly split into a training set and a validation set according to 7∶3.In the training set,the characteristic biomarkers were used to construct a multi-factor logistic regression model and draw a nomogram.Re-ceiver Operating Characteristic(ROC)curve analysis,calibration curve,and Decision Curve Analysis(DCA)were per-formed on the training and validation sets for model verification and evaluation.Results After variable screening and model construction,a multi-factor logistic regression model was obtained,including five variables:gender,aspartate ami-notransferase/alanine aminotransferase ratio(AST/ALT),total cholesterol(TC),superoxide dismutase(SOD),and uric acid(UA).Gender(OR=0.294,95%CI:0.134-0.643,P=0.002),AST/ALT(OR=3.112,95%CI:1.411-6.864,P=0.005),SOD(OR=0.933,95%CI:0.916-0.951,P<0.001),TC(OR=0.564,95%CI:0.383-0.832,P=0.004),UA(OR=0.988,95%CI:0.983-0.993,P<0.001)are the influencing factors of PD occurrence.ROC curve analysis showed that the area under the ROC curve(AUC)of the training set was 0.896(95%CI:0.861-0.930),the AUC of the validation set was 0.853(95%CI:0.787-0.918),and the model classification performance was good.The goodness-of-fit test results of the training and validation sets were P=0.673 and P=0.138 respectively,indicating a reasonable degree of model fit.There is a net gain in the training set and validation set threshold probabilities in the range of 0.05-0.97 and 0.10-1.00,respectively,and the model gain performance is good.Conclusions AST/ALT,TC,SOD,and UA may be potential biomarkers of PD;the auxiliary differential model for PD based on blood markers has certain clinical reference values.