| In today’s society,people are under increasing pressure,and excessive pressure is prone to mental illness.Depression is a common mental illness,and hundreds of millions of people around the world suffer from depression.Depression has a great impact on people’s physical and mental health,and even harms society.Therefore,it is extremely important to find and treat it as early as possible in the stage of depressive tendency.This paper aims to explore a method that can accurately and objectively identify depressive tendency under the premise of ensuring personal privacy,thereby assisting psychologists in judging and analyzing and reducing the probability of misdiagnosis.Therefore,this paper has important theoretical and practical significance for the study of depressive tendency.Although the study of depressive tendency has achieved certain results,there are still deficiencies.First of all,the main method of depressive tendency recognition is mental health self-examination combined with psychological expert consultation.However,most people do not actively seek the help of psychologists for reasons such as privacy protection,and the method is susceptible to subjective factors.Secondly,the study of depressive tendency is based on data from a non-public and imperfect perspective,and the data source is relatively single,unable to lack the multi-faceted study of depressive tendency.Finally,the depressive tendency recognition model is usually adopted a single model,but a single model is prone to problems such as poor generalization ability.In view of the above problems,this paper proposes a method for identifying depressive tendency based on multiple types of features.The main innovations are as follows:1.Using mental health self-examination,eye movement signal and network behavior as a method of emotion recognition,objectively quantify and accurately identify depressive tendency.Firstly,based on the mental health self-examination as a identification method,this paper applies eye tracking technology to depressive tendency recognition,and combines mental health self-examination with eye movement signal to identify depressive tendency.Secondly,personal behavior can identify depressive tendency,and network behavior is an important part of individual behavior,which can identify depressive tendency.This paper combines network behaviors to identify depressive tendency based on mental health self-examination and eye movement signal.2.Combining eye movement features,memory features,cognitive style features and network behavior features,and establishing large-scale and multi-type fusion dataset.Data is one of the key contents of depressive tendency research,because the study of depressive tendency to protect personal privacy,so the data is generally non-public and imperfect,and the data source is relatively simple.In order to make up for the shortcomings of the data,this paper carries out related experimental design and data collection from multiple layers.3.In order to improve the accuracy of the depressive tendency identification method,the depressive tendency recognition model is called the Scanning Stacking Model.Firstly,analysis of the data found that there are complex data relationships,this paper proposes a scanning structure to deal with data relationships.Secondly,this paper builds a two-layer stack structure based on the stacking method.The basic model layer is composed of GBDT,BP neural network,KNN,SVM four models,and the logistic regression model is used as a metamodel.Finally,the scanning structure and the stacking structure are combined to propose a depressive tendency identification method.In addition,this paper objectively and comprehensively evaluates the performance of the model,not only evaluates the performance of each part of the model,but also compares it with multiple indicators from the parameters R square,mean square error,and average absolute error.The results show that the Scanning Stacking Model has the best prediction effect. |
