Development of an Adjustable Pulse Measurement System for Determining the Precise Position for Recording High Wrist Pulse Signals

摘要

Abstract Wrist pulse analysis has been used to examine doshas by applying appropriate force on the radial artery. The pulse signal first increases from low to maximum amplitude with an increased applied force on the radial artery and then decreases as the force increases further. The force where the pulse signal amplitude is maximum is known as the intermediate force level. As per Ayurveda, this level is the most appropriate for pulse signal analysis. Therefore, it is necessary to measure the applied force to record the signal at the intermediate level to get better classification accuracies. However, measurement of the requisite low-force, appropriate calibration models and uncertainty in real-world situations remains challenging. Therefore, this study uses a force-sensing capacitor to measure the force. An experimental setup has been designed to calibrate of sensor output voltage with standard reference static weights. Levenberg Marquardt back propagation artificial neural networks (LMBP-ANN) fitting strategy and traditional method have been proposed for curve fitting. Further, experimental errors and expanded uncertainties of each model have been calculated and compared to select an appropriate calibration model for force measurement. Further, a prototype has been designed that simultaneously measured the wrist pulse signal and downward force from the doshas location. Data were recorded at the intermediate force and superficial levels (25 of an intermediate level) from the Kapha location during pre- and post-meal conditions. The results first show that the LMBP-ANN fitting model shows better accuracies of (99.96) and expended uncertainty of (± 0.25630386) compared to other regression models. Second, the classification accuracies of 80 and 87 at intermediate force and 45.5 and 42.9 at a superficial level (25 of intermediate force level) during pre-meal and post-meal, respectively, demonstrate that pulse signal captured at intermediate level will provide good diagnostic accuracy.