Published in Scientific Papers. Series D. Animal Science, Vol. LXVI, Issue 2
Written by Éva HAJNAL
The main goal is to estimate the HR value from the activity sensor 3D acceleration measurements of the cattle rumen bolus. During the development of the algorithm it was intended to execute the primary calculations on the device's microcontroller and the additional calculations could be performed on the server. The proposed HR estimation algorithm is based on simple data cleaning and peak detection, but the validation and postprocessing of the detection uses AI methods, namely MLP artificial neural network with different cell numbers. The accuracy of the period estimation (IBI) was ±50ms, which means an 8% error. This allows basic alerts to be implemented.
[Read full article] [Citation]
