Estimation of a people’s functional condition after physical activities based on the indicators of the exhaled air condensate registered by polyselective electrochemical sensors with using the projective methods of the multidimensional analysis

This paper connected with Project realized the new principle of noninvasive diagnostics and medicobiological researches with use of the artificial multisensors trained analytical system "Electronic Tonque". Its essence - in application of the massif of polyselective sensors with cross sensitivity for an assessment of the maintenance of physiologically significant components in condensate of the exhaled air, their analysis with use of artificial neurosimilar systems and mathematical methods of processing of big arrays of multidimensional information. The structure is developed and the model of the analytical system which is turning on the sample preparation module for collecting condensate of the exhaled air and three functional modules is created: touch, microprocessor measuring and information, providing formation of images of the studied object, their storing and recognition. Compositions of working solutions are developed for calibration, conditioning, storage of sensors and an assessment of their cross sensitivity, and also a technique of collecting condensate, preparation of sensors for measurements, an assessment of their characteristics in control solutions. Technical and physiological researches which testify to efficiency of the used principle of noninvasive diagnostics of a functional condition of examinees at rest and at the moderate physical activities controlled on indicators of the sensors sensitive to various components of condensate of the exhaled air are conducted. Efficiency of application of a projective method as a principal component analysis for formation of an image of people’s reaction (change of a functional state) to physical activities is shown. The traditional way of creation of a correlation matrix using informative signals array of the examined patients as the training selection. The alternative way of correlation matrix formation based on analogy of two formulas for correlation coefficient and a cosine of the angle between vectors is offered. Thus, coordinates of vectors are sensitivity coefficients (tangents of angle of inclinations of the calibration characteristic) of each polyselective sensor in relation to various test substances. Independence of a choice of test substances provides orthogonality of coordinates of a vector. Thereby, coefficients of correlation are in number equal to a cosine of the angle between the corresponding vectors, and the way of creation of a correlation matrix allows to refuse preliminary "training" of diagnostic system by means of examinees of patients.