J. Weitzenberg and S. Posch (Germany)
Applications, Signal processing, Hidden-Markov-Models, Biosensors
In this paper a Hidden-Markov-Model (HMM) is adapted to analyze curves from amperometric biosensors. The main challenge is to detect a specific time interval during sam pling, representative for the concentration, and to detect a time-point at which measurement may be terminated. This is in stark contrast to other applications of HMMs where they are implemented for pattern classification tasks. An appropriate algorithm for analysis is devised and a specific initialization of the Baum-Welch algorithm developed with a modification of the training algorithm itself. Results for a representative set of signal curve from different analytes are given.
Important Links:
Go Back
