Quantifying arm swing in Parkinson’s disease: a method accounting for arm activities during free-living gait

Table 7 Performance of the classifiers for detecting gait without other arm activities

Group	Sensitivity		Specificity		AUC
Group	LR	RF	LR	RF	LR	RF
Pre-med MAS	0.75 (0.22)	0.72 (0.28)	0.92 (0.04)	0.93 (0.04)	0.93 (0.06)	0.92 (0.10)
Post-med MAS	0.85 (0.19)	0.86 (0.14)	0.90 (0.08)	0.89 (0.09)	0.96 (0.02)	0.96 (0.03)
Pre-med LAS	0.79 (0.17)	0.77 (0.20)	0.91 (0.06)	0.91 (0.10)	0.94 (0.04)	0.92 (0.09)
Post-med LAS	0.80 (0.23)	0.77 (0.28)	0.85 (0.14)	0.84 (0.17)	0.93 (0.08)	0.90 (0.14)

Classifiers include the logistic regression (LR) and the random forest (RF). The PD group is stratified by medication state (prior to (pre-med) and after medication intake (post-med)) and affected side (most affected side (MAS) and least affected side (LAS)). Performance metrics are presented as mean (SD) across participants

ISSN: 1743-0003