Quantification of the Feedback Regulation by Digital Signal Analysis Methods: Application to Blood Pressure Control Efficacy

Abstract

Six different metrics of mutual coupling of simultaneously registered signals representing blood pressure and pulse interval dynamics have been considered. Stress test responses represented by the reaction of the recorded signals to the external input by tilting the body into the upright position have been studied. Additionally, to the conventional metrics like the joint signal coherence Coher and the sensitivity of the pulse intervals response to the blood pressure changes baroreflex sensitivity (BRS), also alternative indicators like the synchronization coefficient Sync and the time delay stability estimate TDS representing the temporal fractions of the analyzed signal records exhibiting rather synchronous dynamics have been determined. In contrast to BRS, that characterizes the intensity of the pulse intervals response to the blood pressure changes during observed feedback responses, both Sync and TDS likely indicate how often such responses are being activated in the first place. The results indicate that in most cases BRS is typically reciprocal to both Sync and TDS suggesting that low intensity of the feedback responses characterized by low BRS is rather compensated by their more frequent activation indicated by higher Sync and TDS. The proposed additional indicators could be complementary for the differential diagnostics of blood pressure regulation efficacy and also lead to a deeper insight into the involved concomitant factors this way also aiming at the improvement of the mathematical models representing the underlying feedback control mechanisms.

Formation of Regression Model for Analysis of Complex Systems Using Methodology of Genetic Algorithms

“…In the case when the number of the input factors is not large, such problems are solved by the so called brute force method [16]. However, when there are a sufficiently large number of the input factors, solution to this problem can be obtained using the machine learning methods, for example, using the genetic algorithms [5,[17][18][19][20][21][22].…”

“…Let us consider the results of analysis for the H-class. According to the algorithm presented in Section I D, on the basis of the data corresponding to the H-class the following regression model was obtained: y 1 (t) = a (1) + a (2) x 5 (t)x 8 (t) + a (3) x 3 (t)x 11 (t) + +a (4) x 3 (t)x 5 (t) + a (5) x 8 (t)x 9 (t) + +a (6) x 11 (t) + a (7) x 1 (t)x 3 (t) + a (8) x 8 (t) + +a (9) x 9 (t) + a (10) x 4 (t)x 6 (t) + +a (11) x 3 (t)x 6 (t) + a (12) x 1 (t)x 8 (t) + +a (13) x 9 (t)x 11 (t) + a (14) x 11 (t) + +a (15) x 6 (t)x 7 (t) + a (16) x 10 (t)x 11 (t) + +a (17) x 1 (t) + a (18) x 6 (t)x 11 (t) + ( 11) +a (19) x 1 (t)x 2 (t) + a (20) x 6 (t)x 9 (t) + +a (21) x 9 (t)x 10 (t) + a (22) x 2 (t)x 3 (t) + +a (23) x 5 (t)x 6 (t) + a (24) x 1 (t)x 7 (t) + +a (25) x 2 (t)x 13 (t) + a (26) x 4 (t)x 11 (t) + +a (27) x 13 (t) + a (28) x 5 (t)x 11 (t) + +a (29) x 4 (t)x 9 (t) + a (30) x 3 (t) + +a (31) x 3 (t)x 8 (t) + a (32) x 9 (t)x 12 (t) + +a (33) x 7 (t) + a (34) x 10 (t)x 13 (t) + +a (35) x 2 (t)x 10 (t) + a (36) x 7 (t)x 13 (t) + +a (37) x 2 (t)x 9 (t) + a (38) x 5 (t)x 13 (t) + +a (39) x 7 (t)x 11 (t) + a (40) x 1 (t)x 5 (t) + +a (41) x 11 (t)x ...…”