Research Of Depression State Recognition Based On Heart Rate Variability

Depression is easy to make patients continue to low mood,insomnia,loss of appetite and other symptoms,serious people will gradually withdraw from society,leading to self-harm and even suicide.Different degrees of depression correspond to different treatment methods,so the early diagnosis and status detection of depression has important practical significance.At present,the diagnosis of depression is faced with the following problems:clinical diagnosis relies on the judgment of scales and psychiatrists,and there is a lack of objective indicators;The mechanism of physiological development between different depressive states is not clear.There is a global shortage of medical resources.Therefore,to explore the performance of physiological indicators in the recognition of depression states,to study the physiological and psychological mechanisms of different depression states,to use the indicators for computer-aided diagnosis,and then to realize the judgment of different depression states,can play a role in the early diagnosis of depression,the selection of treatment programs,reduce medical pressure and other aspects.The physiological changes of mental state can be indirectly reflected in the changes of autonomic nervous system activities.Heart rate variability is closely related to autonomic nervous system,so it is feasible to use heart rate variability to study the state of depression.At the same time,ECG signal has the advantages of non-invasive,easy to collect and high popularization rate of the instrument.In this paper,RR sequences were extracted from ECG signals and analyzed based on the time-domain,frequency-domain and nonlinear characteristics of heart rate variability The physiological change mechanism of different depression states was discussed in depth.A classification model was constructed based on feature matrix to classify depression states.The main research contents of this paper are as follows:(1)Data acquisition and preprocessing.The original signal was preprocessed by wavelet packet transform,and the R points of ECG were identified by the adaptive threshold method,and the RR interval time series was constructed(2)Feature extraction and difference analysis.The characteristics of heart rate variability were extracted from time domain,frequency domain and nonlinear dimension respectively.By extracting the characteristics of heart rate variability,different depression states were quantified.According to the characteristics of different manifestations and their physiological significance,the physiological and psychological response patterns of the human body under different depression states can be inferred.Statistical methods were used to analyze the differences of cardiovascular parameters and heart rate variability between different depression states and healthy controls.(3)Depression status recognition.With the extracted features as input matrix,random forest,support vector machine,precise support vector machine and bagged decision tree classifier were used to realize the recognition of different depression states.