| Background:Military psychological selection is the core issue of military psychology.The current "psychological examination system for young recruits" in our army has achieved good results in effectively ensuring the quality of recruits.However,in recent years,the proportion of recruits without mental disorder with abnormal psychological symptoms has increased significantly.In order to deal with this new situation,the research group carried out an epidemiological survey of high-risk groups of mental disorders in military officers and soldiers,and found that the proportion of high-risk groups of anxiety disorders is the highest,and the identification of high-risk groups of anxiety disorders is the key issue of psychological testing for conscription at this stage.At the same time,although the selfreported personality assessment is the most widely used and ideal test method in military psychological selection,due to the high subjectivity of the self-reported scale,it is difficult to further improve the predictability of psychological test.In order to solve this bottleneck problem,the research group added EEG as an objective index to the existing self-report questionnaire to improve the external gain validity and try to further improve the effectiveness of military psychological selection and evaluation.Objects and methods:From November 7,2019 to November 1,2020,the high-risk group of anxiety disorder,patients with anxiety disorder and normal people were recruited from the Department of medical psychology of PLA General Hospital and the 923 Hospital to participate in the study.The final number of participants was 115,including 39 in high-risk anxiety group,38 in patient group and 38 in normal control group.This study consists of three parts:In the first part,we collected the 3-minute EEG signals of the three groups in the resting state of open eyes and the resting state of closed eyes,to explore the differences of EEG signals among the three groups,and to explore the activation mechanism of the brain regions related to the emotion regulation of anxiety disorder.In the second part,the EEG signals of the military stress response anxiety Prediction Scale were collected,and the EEG difference indexes of the three groups were extracted.In the third part,machine learning technology is used to extract the features of EEG frequency domain indicators and behavioral indicators of the scale and classify the population,so as to improve the recognition ability of anxiety disorder and its high-risk population.Result:1.There were significant differences in the scores of MSAPS and trait anxiety among the three groups(PMSAPS < 43.42,PTA < 54.72).Pairwise comparison showed that the difference of MSAPS score between high-risk anxiety group and normal control group was statistically significant(P < 0.01),between patients and normal control group was statistically significant(P < 0.01),and between high-risk anxiety group and patients was statistically significant(P < 0.05).The difference of trait anxiety dimension between highrisk anxiety group and normal control group was statistically significant(P < 0.01),between patients and normal control group was statistically significant(P < 0.01),and between highrisk anxiety group and patients was statistically significant(P < 0.05).2.In the questionnaire state,the differences of EEG power at the Beta-1 band Cz electrode of the three groups were statistically significant(P < 0.001).By pairwise comparison,the EEG power of the patients were higher than that of the normal control group (P < 0.001),and the EEG power of the high-risk anxiety group was higher than that of the normal control group(P < 0.001).There was no significant difference between the patients and the high-risk anxiety group.3.In the resting state,Delta,Theta and Alpha frequency bands,obvious central line activation appeared in the scalp distribution of EEG power in the three groups.The activation range of the patients was the largest,followed by the high-risk anxiety group,and the lowest in the normal control group.Compared with the normal control group,the highrisk anxiety group and patients had obvious activation of prefrontal lobe.In Beta-1 frequency band,the scalp power distribution of high-risk anxiety group and patients showed a lower degree of central line activation,and the overall power was stronger than that of normal control group.In Beta-2 and Gamma band,the scalp power distribution maps of high-risk anxiety group and patients showed a lower degree of bilateral temporal lobe activation,and the overall power was stronger than that of the normal control group.4.In the resting state with eyes closed,Delta and Theta frequency bands,obvious activation of the central line appeared in the scalp distribution of EEG power in the three groups.The activation range of the patients was the largest,followed by the high-risk anxiety group,and the lowest in the normal control group.Compared with the normal control group,the prefrontal lobe activation was more obvious in the high-risk anxiety group and the patients,and the range and degree of activation in the patients were stronger than those in the high-risk anxiety group.In Alpha band,obvious activation of central line appeared in the scalp distribution of EEG power in three groups.The central line in highrisk anxiety group had the largest activation range and left occipital lobe activation,followed by that in patients but no occipital lobe activation,while that in normal control group had the lowest activation degree but bilateral occipital lobe activation.In Beta-1 frequency band,obvious activation of the central line appeared in the scalp distribution of EEG power in three groups,with the largest activation range in the patients,followed by the high-risk anxiety group,and the lowest in the normal control group.5.In the questionnaire response state,in Delta,Theta and Alpha frequency bands,the distribution of scalp distribution of EEG power in the three groups was roughly the same,with obvious activation of prefrontal lobe and central line,and the degree and range of activation decreased with the increase of frequency.The patients had the highest degree of activation and occipital lobe activation,followed by high-risk anxiety group,and the normal control group had the lowest degree of activation.In Beta-1,Beta-2 and Gamma bands,the scalp distribution of EEG power in the patients and the high-risk anxiety group was similar,and obvious activation of prefrontal lobe and occipital lobe was found in both groups,which was significantly different from that in the normal control group.In Gamma band,bilateral temporal lobe activation was found in patients and high-risk anxiety group.6.The sensitivity and specificity of EEG Beta-1 power band at Cz electrode were 77.9% and 76.3%,respectively.7.SVM classifier was used to classify the patients,anxiety high risk group and normal control group.The results showed that the number of subjects in the above three groups was based on the above three groups × Band power value × The results show that the accuracy of group prediction for unknown individuals can reach 76%.SVM classifier was used to classify the patients,anxiety high risk group and normal control group.The results showed that the number of subjects in the above three groups was based on the above three groups × w PLI value × The results show that the accuracy of group prediction can reach 81% for unknown individuals by using KNN classifier.Conclusion:1.the study verified the difference between the high-risk group of anxiety disorder and the normal group and the patients with anxiety disorder.It was found that there was gradient difference in symptom and behavior,but there was no significant difference between the high-risk group of anxiety disorder and those with anxiety disorder in electrophysiological index.2.the beta-1 band EEG power at Cz electrode has potential to be a characteristic electrophysiological index for screening anxiety disorder patients and high-risk groups under the response anxiety prediction scale of military stress.3.the military stress response anxiety prediction scale has both experimental properties as EEG stimulation materials and prediction attributes as measuring tools.Through EEG based measurement,subjective behavioral indicators and objective electrophysiological indicators can be obtained simultaneously,and indicators can be integrated by using machine learning technology,the ability of identifying anxiety disorder patients and their high-risk groups is improved to the maximum extent. |
PDF Full Text Request