Assessment of Gait Symmetry and Gait Normality Using Inertial Sensors: In-Lab and In-Situ Evaluation - Biomedical Engineering Systems and Technologies - page 223

Ta b l e 1 . Correlation of MAP and GPS val-
ues with the proposed normality measure
(waist sensor node)

Ta b l e 2 . Correlation of MAP-symmetry and
GPS-symmetry values with the proposed
symmetry measure (shank sensor nodes)

sensor accelerometer
placement waist
histogram transition
no. symbols 18
variable r p-value
MAP knee flex. 0.77 < 0.0001
MAP hip flex. 0.82 < 0.0001
MAP pelv. rot. 0.71 < 0.0001
GPS

sensor gyroscope

placement shank

histogram symbol period

no. symbols 20

variable r p-value

ankle flex. 0.64 < 0.0001

knee flex. 0.81 < 0.0001

hip flex. 0.68 < 0.0001

0.81 < 0.0001 all 0.84 < 0.0001

30

25

20

before surgery
after follow-up

15

10

5

0

mobility self-care activities pain/disc. anxiety/depr.

Fig. 4. Questionnaire results from before the surgery and after the follow-up sessions. Lower
scores correspond to more patients in better health

walking aid is shown in Figure 6. There is a clear difference between the symmetry of
patients using two crutches at baseline and patients walking with no aid at follow-up.
However, none of the distributions were significantly different.
Normality results are shown in Figure 7, measurements for each patient were av-
eraged over both trials of each session. Similarly, the normality index ranges between
0 and 100, and a low value indicates good normality. In this case, the follow-up mea-
surements were better than baseline measurements for all patients. A Wilcoxon test
indicated that baseline and follow-up groups were statistically significantly different,
p < 0.0001. Figure 8 illustrates the distribution of the normality index according to walk-
ing aid. As expected, the normality index for those patients walking without aid was,
on average, better than the others. A Kruskal-Wallis test indicated that the free walking
group was statistically different from the walker and crutches groups, and that the one
crutch group was statistically different from the walker group.
In order to calculate the correlation between normality and walking aid, each cate-
gory was represented by a number. In the order shown in Figure 8, walker was