Time Series Analysis In Treatment For Depression

The main purpose of thesis study is to find the evolution trend of depression and anxiety level of depression patient by ARIMA model and exponential smoothing, to understand the dynamic autocorrelation of the variables by ARIMA model, and to forecast the variation of the future condition. Moreover, it supposed to established VAR model to test patient's depression severity and anxiety level. Granger causality test has been used to analyses the causality between these two variables.This study has tracked 6 inpatients of depression in NanJing Brain Hospital for 3 months, 4 of who withdrew in the process. The remaining patients were asked to complete Self-Rating depression Scale and Self-Rating Anxiety Scale everyday by the means of self-reporting. Furthermore, the author uses interview method to ensure reality of information reported by the patients, and ensured the reality of information with interview method, and investigates the social support status of patients via Social Support Scale. Time series analysis is conducted on daily depression and anxiety data using Eviews5.0.The main results are indicate:1. Apparently, depression severity of two patients decreased in the study, but the fluctuating range of depression symptom patients in two adjacent days are inconsistent. Different depression types of patients are considered to be the explanation of this phenomenon. For the different depression type of patients, the variety of single episode depression patient is relatively slow, and the depression state fluctuation of recurrent depression patient is more distinct than the former.2. The depression severity of two patients in the same day is restricted by the state of the previous 5 days, depression severity of the day before is most affective factor for the very day, and there are differences in anxiety level restricted by previous state of anxiety between two patients.3. The current and long-term trend of depression severity variety in the self-report of patients is basically the same, but that of the anxiety level is inconsistent.4. The depression severity and anxiety level of patient LY could forecast action for each other, comorbidity phenomenon did occurred in the study, and the anxiety level of patient ZSB could forecast action for the depression severity only.Time series analysis provided a new scientific means to understand and predict human psychology and behavior for us, it offered new clue for the traditional studies of clinical treatment for depression as well. The application value of time series analysis presented in this thesis lies in determining the change law of the depression and anxiety of patients, and making the dynamic forecasting become possible. Time series analysis is also helpful to control the condition of the patients, even the mutual forecast relation between depression and anxiety.