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

Table 5 Performance of the classifiers for gait detection

Group	Sensitivity		Specificity		AUC
Group	LR	RF	LR	RF	LR	RF
PD pre-med	0.83 (0.10)	0.89 (0.09)	0.96 (0.03)	0.95 (0.03)	0.98 (0.02)	0.98 (0.01)
PD post-med	0.88 (0.09)	0.95 (0.05)	0.94 (0.04)	0.91 (0.05)	0.98 (0.01)	0.99 (0.01)
Controls	0.92 (0.03)	0.96 (0.02)	0.88 (0.06)	0.82 (0.09)	0.97 (0.01)	0.98 (0.01)

Classifiers include the logistic regression (LR) and random forest (RF). Performance is stratified by pre-medication (pre-med) and post-medication (post-med) states in the PD group. Results are shown for the most affected side of PD participants, and for controls, matching of affected side stratifications based on self-reported hand dominance was applied. Performance metrics are presented as mean (SD) across participants

ISSN: 1743-0003