Extraction and selection of parameters for evaluation of breath alcohol measurement with an electronic nose

The ethanol concentration in realistic breath samples was analyzed using an electronic nose. Conditions were selected so that the samples would reflec t those collected in a real drunk driver situation. Hence, parameters such as intake of food and beverage, tobacco habits, as well as the order of par ticipating volunteers were allowed to be variable. The setup was unexpected ly robust towards inter- and intrapersonal variations in breath samples as well as long-term variations. The standard error (16 mol ppm) was the limit ing factor but the statistical detection limit was well below 0.1 ppm. The standard error corresponds to between 9% (Austria) and 36% (Sweden) of lowe st legally accepted levels. Even though this is regarded as a significant e rror, there are several options of optimization. Incorporating feature extr action and forward selection together with artificial neural network for pr ediction of the ethanol concentration showed, besides increasing the accura cy, to be a valuable tool generating feedback of possible improvements of t he sensor array. (C) 2000 Elsevier Science S.A. All rights reserved.