4 Results and Discussion
For the investigations, physiological and activity profile data were captured from two
healthy participants over five repetitions of both activity scenarios, where each
activity scenario consisted of three activities. Although data from a total of 60
activities were recorded, only 83.33% of the datasets contained complete information,
due to issues with the Bluetooth communications between the wearable sensor
devices and the receiving computer. Initially, detailed results from one participant for
a single performance on one activity will be given, along with the corresponding
analysis. This will be followed by selected results and analysis from a single
participant on the remaining set of activities performed.
Figure 2 illustrates the set of values for the heart rate, as determined using the R-
peak detection method previously described in Section 3.3, alongside the
corresponding set of values obtained for the Mean Square Error (MSE) during
CUSUM analysis. The activity profile information captured for the activity, in terms
of the sets of normalized acceleration values from both the upper-body and lower-
body accelerometers, are subsequently presented in Figure 3.
Fig. 2. Activity Sit Down : Heart Rate & Mean Square Error (MSE). CUSUM result is given in
both graphs ( dashed vertical line ).
From the CUSUM analysis, it has been discovered that there is a significant change in
the trend associated with the heart rate values during the activity. The corresponding
change point of the trend is identified during CUSUM analysis by minimizing the total
individual variances for the data points within the two segments separated by the change
point. In Figure 2 the position of the change point is illustrated with a dashed vertical line