Algorithms and strategies for extracting optimal information from chemical sensing systems

The output signals of chemical sensing systems, i.e. of sensors used to detect chemical quantities, typically consist of a complex superimposition of three different contributions: useful information, non relevant (but systematic) variations, and noise. For an efficient extraction of the highest possible amount of useful information, the application of multivariate methods is definitely more effective than commonly used univariate approaches. However, multivariate methods themselves could not allow the extraction of the whole information content of interest. The goal may be achieved by an efficient use of additional strategies, suitable to consider other aspects such as signal shape, time-evolution of a given sensor response or interactions among signals measured with different sensors. The performance of the sensor(s) is improved and the final output may consist of an optimized set of parameter values.