Information Technology Reference
In-Depth Information
change in finger temperature signal can easily be captured by calculating the slopes
for each step.
Calculating the Mean and Standard Deviation.
Our opinion is that neither the
mean value nor the standard deviation of the FT measurement was adequate indica-
tors of stress. For instance, consider two signals: one is increasing from 20° C to
30° C, the other decreasing from 30° C to 20° C. Although both signals have the same
mean/standard deviation value in the duration, however they indicate opposite stress
levels. An example of such measurement is illustrated in Fig. 19.
Fig. 19.
Temperature is increasing and decreasing
From Fig. 19, it can be seen that temperature is increasing and decreasing from
27° C to 29° C, so one measurement might be classified as stress and other one as
relaxation. If we consider the mean temperature or standard deviation for both the
cases the value will be the same. Alternatively, the mean slope value might be a feasi-
ble feature to convey a relation with stress. If the mean slope value is sufficiently
positive, it will be a clear indication that activity of the sympathetic nervous system is
decreasing e.g. relaxation, otherwise an indication of stress. But if the mean slope is
around zero, it shows a situation with greater uncertainty in diagnosing stress and a
greater probability in making a poorly justified decision.
5 Feature Extraction and Case-Based System
It is often convenient to classify a signal if it can be simplified to a form that describes
the characteristics of the signal i.e. extract features that represent the signal. During
diagnosis, when performed manually, an experienced clinician often classifies FT
signals without intentionally pointing out all the features used in the classification.
However, extracting appropriate features is of great importance in performing
accurate classification in a CBR system.
Search WWH ::
Custom Search