Motor task complexity matters more in dual task gait performance for early-stage Parkinson's disease identification.

Parkinson's disease (PD) patients exhibit more impaired gait performance during motor cognitive dual-task tests. However, the effects of motor and cognitive task complexity on early PD identification have not been systematically investigated. 25 idiopathic early PD patients and 14 age-matched healthy elders were recruited and required to perform a complex motor task with/ without three different cognitive tasks. Complex task interference features were respectively calculated based on gait spatiotemporal parameters and joint kinematics. The effects of task complexity and group on CTIs were firstly investigated using a repeated measures analysis of variance. A support vector machine-based PD classifiers were built based on distinguished CTI features. The results demonstrated that the motor task complexity had a more significant effect on gait performance which contributed to significantly improved accuracy in PD classification. The best classification was achieved at turns with backward recall task (accuracy: 97.7%; precision: 98.9%; recall: 97.7%). The study proved the great potential application of a turning-based dual-task test paradigm in clinic for identifying early-stage PD.