Diagnostic value of AI-based motion assessment in differentiation of Parkinson's disease and multiple system atrophy-Parkinsonian type
Wang Qianyao
Ren Na
Chen Jilin
Li Hua
Li Min
Zhang Shufeng
Yu Jin
Qian Hairong
Abstract：Objective To use AI-assisted motor dysfunction assessment for quantitative evaluation of motor function in Parkinson's disease(PD)and multiple system atrophy-Parkinsonian type(MSA-P)in order to achieve accurate differential diagnosis.Methods A total of 105 participants aged ≥60 years were consecutively enrolled from the First and Third Medical Centers of Chinese PLA General Hospital between January and September 2024.Based on diagnostic criteria,they were divided into a PD group(48 cases),a MSA-P group(31 cases),and a control group(26 cases).The general information was collected,and the motor function was evaluated with Move-ment Dysfunction Assessment Software in order to assess the diagnostic value of the AI-assisted assessment in differentiating between PD and MSA-P.Results Significantly differences were observed among the three groups in terms of facial expression indicators,bilateral finger tapping frequency,bilateral hand movement frequency,right hand movement amplitude change rate,bilat-eral palm flipping frequency,bilateral toe tapping frequency,freezing load of bilateral toe tapping,bilateral leg flexibility frequency,right leg flexibility amplitude change rate,freezing load of bilat-eral leg flexibility,upright extension angular velocity,turnaround time,forward step frequency,backward step frequency,forward average stride length,backward average stride length,forward average walking speed,backward average walking speed,forward average step width,backward average step width,bilateral postural tremor frequency,bilateral postural tremor maximum am-plitude,bilateral action tremor frequency,bilateral action tremor maximum amplitude,and com-parison of bilateral resting tremor frequency(P＜0.05,P＜0.01).The MSA-P group exhibited significantly lower blink frequency,maximum amplitude and frequency of facial tremors,upright extension angular velocity,and step frequency,while higher ratio of mouth opening duration and longer turnaround time when compared with the PD group(P＜0.05,P＜0.01).The AUC value of the combined nine motor function indicators and the five facial expression indicators in differ-entiating PD from MSA-P was 0.943(95％CI:0.895-0.991,P=0.000)and 0.925(95％CI:0.870-0.981,P=0.000),respectively,both better than that of individual indicators.Conclusion Combi-nation assessment of facial expression,posture,gait with AI assistance can contribute to the dif-ferential diagnosis of PD and MSA-P.
Keywords：Parkinson diseasemultiple system atrophyartificial intelligencediagnosisdifferential
Publication Date：2025-04-15
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 482-487 )