Abstract
Parkinson’s disease (PD) patients usually develop facial abnormalities as a result of both motor and nonmotor
impairments. In the present study, 96 healthy Japanese control subjects and 97 Japanese PD patients were
analyzed for the facial appearance age, emotion and skin condition by novel artificial intelligence (AI) softwares.
Because an appearance age of our Asian subjects was younger than their real age, we newly adjusted it for
Japanese. The age-adjusted AI software looked age of PD patients as 2.4 years older than healthy control
subjects (*p under 0.05). The gap between appearance age and real age was larger in male and younger PD patients
than in female and older patients, respectively. In the AI facial emotion analysis, expressionless was obviously
more frequent and happiness was less frequent in PD than in control (*p under 0.05). On the other hand, facial skin
condition didn’t show any difference between PD and control. The present AI face recognition software thus
clearly detected the facial outlook of PD corresponding to clinical expressions.
