A.N. Kalinichenko - Dr.Sc. (Eng.), Senior Research Scientist, Department of Biotechnical Systems, Saint-Petersburg State Electrotechnical University «LETI» n. a. V. I. Ulyanov (Lenin). E-mail: ank-bs@yandex.ru
A.P. Nemirko - Dr.Sc. (Eng.), Professor, Department of biotechnical system, Saint-Petersburg State Electrotechnical University «LETI» n. a. V. I. Ulyanov (Lenin). E-mail: apn-bs@yandex.ru
I.S. Salamonova - Post-graduate Student, Department of Biotechnical Systems, Saint-Petersburg State Electrotechnical University «LETI» n. a. V. I. Ulyanov (Lenin). E-mail: i.salamonova@yandex.ru

One of the most important functions of mechanical ventilators is measurement of the lungs complians and the airways resistance. Using these two parameters together with scalar flow, pressure and volume signals and also the loops "volume-pressure" and "flow-volume" an anesthesiologist or resuscitator can assess current condition of a patient under mechanical ventilation. This data considerably reduce a risk of the patient condition complications. Some difficulties take place in the measurement of complians and resistance connected with the signals instability, noises and the patient-s spontaneous breath. The following three methods are used to suppress noises: averaging, filtering and polynomial approximation. Experiments show that the two first approaches do not provide efficient suppression of noises. So the polynomial approximation technique was proposed for use. This paper presents the results of experimental examination of some parameters of the polynomial approximation: the polynomial order, the approximated fragment duration, the step of the fragment move. The most optimal set of the proposed procedure parameters that provides the best accuracy and stability of results was determined with the use of real recordings of flow, pressure and volume signals. Mean square error between the estimated complians and resistance values and those obtained from a control device was used as a criterion for the parameters optimization. The presented investigation shows advantage of the approach based on combination of inspiration cycles moving averaging and polynomial approximation.