Emotion Recognition Of Individual Depression Patient Based On Finger Pulse Signal

In March 2018,data released by the World Health Organization(WHO)showed that more than 300 million people worldwide are suffering from depression.Depression is a common and easily overlooked mental illness,brought great mental pain to the patients and their families.At present,the diagnosis of depression mainly relies on the Mental Disorder Predictive Scale,which lacks objective biological indicators.Moreover,the functional near-infrared spectroscopy(fNIRS)based on brain imaging is a new technology for the diagnosis of depression.The necessary equipment is expensive and difficult to carry,and patients with depression cannot be monitored in time.With the continuous development of information technology and computer science,the traditional method of diagnosing depression based on symptomology of diseases cannot meet the demand for more systematic and in-depth research on depression.The rising technology called Computational Psychiatry introduces computing and statistical method explores the internal pathological mechanism of depression.Computational Psychiatry uses high-dimensional complex data to further reveal the process of information processing and assist in the identification of various mental diseases,which contributes to the onset,mechanism,prevention,diagnosis and treatment of mental diseases.Pulse,as an important physical parameter of the human body,contains a large amount of physiological and spiritual information from human body.Its collection equipment is low-cost and easy to carry.The collection method is relatively simple."Pulse Diagnosis" has unique advantages in theoretical research and clinical diagnosis of various diseases,and has been widely used in traditional Chinese medicine in China.Therefore,applying the pulse signal to monitoring the emotional state of patients with depression to assist in the diagnosis and treatment of depression has broad prospects for development and application space.This paper aims to process or extract features from finger pulse signals of individual depression patient,and use machine learning technology to identify and analyze the mental state of patients with depression based on the processed finger pulse signals or extracted finger pulse features.The main research results are as follows:1.Construction of finger pulse signal data set:the sampling collection rules and calibration rules of finger pulse signal are determined.The finger pulse signal is de-noising processed and segmented into a separate pulse cycle by detecting the peak value of it.The construction rules of finger pulse signal data set are determined.2.The deep learning technology is applied to the task of mental state recognition based on finger pulse signal:the two-dimensional ResNet is transformed into one-dimensional ResNet,so that it can recognize mental state according to finger pulse signal.By adjusting the scale of one-dimensional ResNet model and using large convolution kernel strategy,the recognition performance of the model is further improved.3.Mental state recognition based on finger pulse feature and fully connected neural network:The paper designed finger pulse feature extraction rules and methods,and constructed finger pulse feature data set.Combining the advantages of neural network and traditional machine learning algorithm,neural network is used to recognize mental state according to the features extracted by artificial rules.A neural network composed of full connection layer is built for mental state recognition according to finger pulse features.Through analysis and experimental verification,the neural network composed of full connection layer is effectively improved,and the influence of each feature on recognition effect is statistically analyzed.