Fig. 2
From: Wearable sensor shown to specifically quantify pruritic behaviors in dogs
The importance of high frequency data sampling. Schematic of data sampled at low frequency misrepresents the original sinusoidal curve (left). Conversely, high frequency sampling enables an accurate digital representation of the original signal – facilitating capture of the necessary data fingerprint. Applied to actual data for dog behaviors scratching and running (right), data points sampled at high frequency are shown in blue, while the subset that would represent low frequency sampling is shown in red. The upper line graphs in blue connect the high frequency samples, and show a significant amount of detail from the captured signal. The lower line graphs connect the low frequency samples in red, and show far less detail making scratching and running examples not easily distinguishable. The low frequency sampling of the behaviors does not enable the accurate digital recreation of the signal, due to the loss of the high frequency components