Classification Framework for Electropulsegraph Waves

Abstract

Electropulsegraphy is a medical device that was invented by an orient medical physician and a few engineers to help the physicians to diagnose patients in more systematic way by analyzing waves generated from the device. With the use of the device, doctors can diagnose patients based on charts rather than by feeling purses manually as in traditional oriental medicine practice. The device uses traditional microphones as sensors attached to particular locations on a body of a patient (e.g., a wrist) and generates a number of waves that reflect the physical states of the patient. The device has been used for several decades by physicians in Korea, and it undergoes functional upgrades both in hardware and software aspects recently. As one of those upgrading efforts, we strive to make the diagnostic process automatically by applying the well-known machine learning algorithm-logistic regression. In this article, we provide a framework of the preliminary study with experimental classification results. Training data sets collected for decades are used to estimate the parameters of the logistic regression. And the parameters are used to classify wave inputs chosen at random.