Anxiety Disorder In High-Risk Populations:An Eye Movement Study Based On Self-Reported Questionnaires

BackgroundSince ancient times,regardless of how the form of war changes,“people” have consistently been the determinants in the outcome of wars.As special occupational groups,soldiers have always been under enormous psychological pressure because of their closed living environment,strict discipline requirements,high training intensity,high-risk factors,and physical loss.Consequently,mental health problems are prone to occur in individuals in a military setting.Therefore,the scientific,efficient,and psychological selection of military personnel is a practical and crucial problem encountered in military psychology in all countries.Our military service system,which mainly combines voluntary and compulsory military service,determines that the recruitment work of the army is characterized by the number of recruits,relatively concentrated personnel,high time density,and comprehensive testing content requirements.The form of centralized organization within a limited time limits the choice of psychological evaluation methods.Therefore,the army’s psychological selection has continually been based on self-reported questionnaires supplemented by structured interviews.As the most mainstream psychological evaluation method presently,self-reported questionnaires are widely used in the psychological selection of military personnel in various countries and have achieved positive results in the actual testing process.However,it has inherent limitations;participants are highly subjective when answering the questionnaires and are easily affected by social desirability and cultural factors.Additionally,they even falsify to achieve a certain social purpose,which leads to a certain response bias and reduces the reliability and accuracy of the test results to a certain extent.Difficult and unavoidable subjectivity is a critical challenge that needs to be overcome in the current psychological assessment field.With the rapid development of science and technology,such as eye-tracking,near-infrared,and facial micro-expression capture,militaries in various countries have begun using cognitive neuroscience technology to optimize and compensate for the defects of traditional self-report questionnaires.In this context,the National Recruitment Psychological Testing Technology Center,where this research group is located,proposed the research concept“consciousness-cognition”outer loop approach in 2014 and established the simultaneous collection of eye-tracking and EEG data based on self-reported questionnaire responses.The multimode fusion technology of cognitive neuro-objective indicators,such as facial micro-expressions,has highlighted the research and development of China’s new generation of recruitment psychological detection systems.Anxiety disorders are a group of mental disorders with anxiety syndrome regarded as the main clinical manifestation,with other clinical manifestations including emotional arousal,excessive worry,and increased autonomic nervous system activity.According to the latest epidemiological survey report on mental disorders of high-risk groups in the army,the prevalence of anxiety disorders in various military services are 6.55%in the army,7.02%in the navy,8.31%in the air force,and 7.44%in the rocket army,all of which are at high risk of mental disorders in all military services.Early screening of anxiety disorders is vital in forming and maintaining military combat training,security,stability,and effectiveness.Improving the detection rate of anxiety disorders based on existing self-report questionnaires has always been one of the key issues of this research group.In previous research,this research group applied EEG,eye-tracking,facial micro-expression,facial circulatory blood flow,and other technologies to a multi-integration study of individuals with schizophrenia,antisocial personality disorder,and other mental disorders and some other high-risk groups.Consequently,we obtained the corresponding research results.This study considered anxiety disorders and high-risk groups as the main research objects.Eye-tracking,an experimental technique that reflects psychological activities from the perspective of visual characteristics,was used as positive and negative controls,to identify the eye-tracking characteristics of anxiety disorders and their high-risk groups when answering self-reported questionnaires.Hence,this enables early identification of anxiety disorders and high-risk groups and provides theoretical support and a practical basis for the realization of multi-index integration and the development of a new generation of recruitment psychological testing systems.MethodsThis study was divided into two parts.Patients diagnosed with anxiety disorders were regarded as the anxiety disorder group.Patients with anxiety disorder-prone traits and some symptoms of anxiety disorders,but not enough to be diagnosed with anxiety disorders,were regarded as the high-risk anxiety disorder group,which was based on the samples of normal/healthy people.The first part adopts the multi-quality fusion research paradigm of“self-report questionnaire + eye tracking” to explore the differences in the scores and basic eye-tracking characteristics of the anxiety disorder,anxiety disorder high-risk,and normal contrasting groups in response to anxiety-related questionnaires.Fusion research provides data support,a theoretical approach,and a basis for the next step of machine learning and crowd classification.The second part uses a machine-learning method to perform pattern recognition on the extracted eye-tracking features.The multimode fusion classification indicators of the “self-reported questionnaire + gaze tracking” of the three groups of people were extracted and analyzed using the random forest algorithm and the XGBoost algorithm as classification algorithms.The test of population classification was carried out to provide practical examples and index bases for the screening of anxiety disorders and high-risk groups in the new generation of recruitment psychological testing systems based on multimode fusion technology.ResultsThe first part compares and analyzes the results of the “Military Stress Reactive Anxiety Prediction Scale” in the anxiety disorder group(98 cases),the high-risk group with anxiety disorder(99 cases),and the normal group(99 cases)and their simultaneous collection.Eye-tracking characteristics and four main outcomes were identified.Result 1: There were differences in the responses among the anxiety disorder,anxiety disorder high-risk,and normal group(F=119.165,P<0.001).After the comparison,it was found that the anxiety disorder group’s score was significantly higher than that of the highrisk anxiety disorder and normal groups(P<0.001).Moreover,the high-risk anxiety disorder group’s score was significantly higher than that of the normal group(P<0.001).This outcome confirms the effectiveness of the current self-report questionnaire as the main psychological assessment tool;further,it proves the rationality and effectiveness of this group’s strict control over the experimental population.Result 2: The anxiety disorder,anxiety disorder high-risk,and normal groups had differences in the total number of fixation points,total fixation time,number of retrograde gazes,and pupil diameter;additionally,there were differences among the groups(P<0.001 or P<0.05).There was no significant difference between the anxiety disorder and the highrisk anxiety disorder groups;however,there were significant differences compared with the normal group.Specifically,for each index,for the total number of fixation points and the number of retrospectives,the anxiety disorder group had the largest number,which was significantly higher than the anxiety disorder high-risk and normal groups(P<0.001);the anxiety disorder high-risk group was second,followed by the normal population(P<0.05).The differences among groups were statistically significant.Regarding the index of total fixation time,the anxiety disorder group had the longest time,which was significantly longer than the anxiety disorder high-risk and normal groups(P<0.001);however,there was no statistical difference between the anxiety disorder high-risk and normal groups.For the pupil diameter index,the order was anxiety disorder group>anxiety disorder high-risk group>normal group,and the post-hoc comparison results showed a statistical difference among the three groups(P<0.001 or P<0.05).This outcome suggests that anxiety disorders and their high-risk groups may have cognitive processing characteristics,such as impaired attentional resource allocation,decreased processing speed,and longer information processing time when answering self-reported questionnaires.The second part is based on the research in the first part,adopts the machine learning method,takes the questionnaire score,eye-tracking index,and the combination of the questionnaire score and eye-tracking index as the inclusion index of the machine learning classifier,and explores the independent self-reported questionnaire.Based on the evaluation,the main results of the feasibility of adding an eye-tracking modality to improve the classification and recognition efficiency of anxiety disorder patients,high-risk groups of anxiety disorder,and normal people are as follows: the classification and recognition accuracy rates of the machine learning model based on the questionnaire score,the eyetracking index,and the combination of the questionnaire score and eye-tracking index for the three groups are 66.2%,49.3%,and 74.7%,respectively.“Questionnaire” eye-tracking indicates that adding eye-tracking modalities based on questionnaire scores can further improve the recognition efficiency of the machine learning model,which is improved by8.5%.ConclusionThis study yielded two main results.The first one is to study the different eye-tracking characteristics of anxiety disorders and their high-risk groups compared with normal people when they answer self-reported questionnaires,while the second is to use machine learning.The method initially established a classification index system with a certain recognition rate and provided a practical example of its possible application in the psychological testing of conscription.