Minding your steps: a cross-sectional pilot study using foot-worn inertial sensors and dual-task gait analysis to assess the cognitive status of older adults with mobility limitations

Table 6 Classification performance based on logistic regression and ROC Analysis (n=29)

Gait metric	AUC	SE	p-value	Sens./Spec.
Stride time DTE (%)	0.740	0.097	0.028*	0.813 / 0.692
Foot flat ratio DTE (%)	0.760\(^\dagger\)	0.090	0.018*	0.625 / 0.846
Pushing ratio DTE (%)	0.764\(^\dagger\)	0.092	0.016*	0.750 / 0.769
Speed DTE (%)	0.707	0.096	0.059	0.625 / 0.692
Liftoff angle DTE (%)	0.736	0.094	0.032*	0.688 / 0.769
Min toe clearance DTE (%)	0.716	0.097	0.048*	0.750 / 0.615
Heel 3D path length DTE (%)	0.712	0.097	0.054	0.625 / 0.769
DT Stride time variability (%)	0.755\(^\dagger\)	0.095	0.020*	0.750 / 0.769
DT Stride length variability (%)	0.760\(^\dagger\)	0.101	0.018*	0.750 / 0.769
DT Speed variability (%)	0.731	0.101	0.035*	0.750 / 0.692
DT Liftoff angle variability (%)	0.721	0.098	0.044*	0.688 / 0.846
DT Heel trajectory forward MAD (mm)	0.745	0.097	0.025*	0.813 / 0.769
DT Toe trajectory forward MAD (mm)	0.745	0.100	0.025*	0.813 / 0.769
DT Heel 3D path length variability (%)	0.716	0.100	0.048*	0.750 / 0.692
DT Toe 3D path length variability (%)	0.745	0.098	0.025*	0.750 / 0.692
DT Loading ratio asymmetry (%)	0.716	0.102	0.048*	0.938 / 0.538
DT Liftoff angle asymmetry (%)	0.726	0.095	0.039*	0.750 / 0.692

Sensitivity (Sens.) and Specificity (Spec.) were obtained for the optimal cut-point based on ROC curve analysis. p-values under the null hypothesis that AUC equals 0.5
AUC area under the curve, ROC receiver-operating characteristic, SE standard error (under nonparametric assumption), DTE dual-task effect, DT dual-task, MAD mean absolute deviation
*\(p<0.05\), bold values indicating significance
\(^\dagger AUC > 0.75\)

ISSN: 1471-2318