Nonlinear processing of wrist pulse signals to distinguish diabetic and non-diabetic subjects

Abstract

In pulse diagnosis, the pulse signals obtained at wrist have been used for analysis of certain diseases in ancient systems of medicine in which the practitioner feels the pulse of the subject by placing his three fingers on the subject’s wrist at three distinct radial pulse point locations. The preliminary studies show that there are many conventional linear techniques applied to analyze the wrist pulse signals and less focus on non-linear techniques. Hence, the main aim of this research is to apply Recurrence Plot and Recurrence Quantification Analysis (RQA), a nonlinear technique to analyze the wrist pulse signals for distinguishing between diabetic and non-diabetic subjects. Wrist pulse signals from 32 subjects were recorded during the morning hours and were analyzed using RQA techniques. The results show significant differences in the RQA parameters of the wrist pulse signals as they are obtained from the recurrences occurring in the phase space plots of the wrist pulse signals. It was found that parameters like entropy, divergence and average diagonal line length showed significant variations for diabetic and non-diabetic subjects. Therefore, it can be concluded that RQA parameters can be used effectively to identify diabetic and non-diabetic subjects and thus may be applied on the wrist pulse signals for early detection of various diseases.