| Background:In recent years,military suicides have shown an increasing trend with the rise in the number of suicides worldwide.Military suicides can lead to noncombat attrition in the force,a decrease in military combat effectiveness,and a serious threat to the physical and mental health of officers and soldiers.Suicide prevention in the military has always been a focus of public health efforts,however,the current primary screening instrument based on the suicide tendency self-statement inventory has hindered prevention efforts.First,judging subjects solely on the results of their subjective responses leads to lower rates of accurate reporting.Second,the high face validity of the existing suicidal tendency scale is likely to alert subjects,further leading to increased concealing and underreporting.Therefore,improvement of the existing screening instruments is an urgent need for suicide prevention.Based on this,we used metaphorical language form to reduce the surface validity of the screening material and combined with eye-movement technology to record objective indicators when subjects responded,thus compensating for the drawbacks of the existing self-report method of suicidal tendency screening.According to the properties of metaphorical language,using metaphorical suicidal tendencies corpus as an arousal cue for suicidal tendency people has better effect,and metaphorical language was used as specific linguistic stimulus in this study.The uncontrolled eye movement trajectory can reflect the attentional bias of mental activity and then infer cognitive processing,which can increase the objectivity and accuracy of suicidal tendency screening to a certain extent,so the eye-movement technique was used as the technique used in this study.In summary,this study takes suicidal tendency as an entry point,using a metaphorical suicidal tendencies corpus as specific linguistic input information and thus exploring the information of eye movement characteristics of people with and without suicidal tendency under the same linguistic stimuli using eye-movement technology,providing a new practical basis for improving the existing suicidal tendencies screening tests.Methods:Based on the presence or absence of suicidal tendency,two groups of subjects were recruited in this experiment,the suicidal tendency group(experimental group)and the non-suicidal tendency group(control group,both referred to as control group hereafter).There were no significant differences in age,gender,occupation and education between the two groups of subjects.The study was divided into three parts.Part I: A qualitative study to obtain a corpus of metaphorical soldiers’ suicidal tendencies.This part of the study used in-depth interviews to collect the metaphorical corpus of suicidal tendency soldiers.According to the interpersonal theory of suicide proposed by Joiner,psychological interviews were conducted with the subject group to collect their experiences about suicide and their metaphorical expressions during the new training period,and innovative forms of metaphorical language expressions were used in order to obtain clinical information about the target population and to improve the validity and accuracy of metaphorical corpus acquisition.Part II: An eye-movement difference analysis study between the suicidal tendency group and the control group on the metaphorical corpus and the high surface validity corpus.This study consisted of two parts: the first part of the study used expert evaluation and online language collection to construct a high surface validity suicidal corpus as a control material for the metaphorical suicidal corpus;the second part of the study combined eye-movement technology to collect and compare the differences in eye-movement characteristics between the suicidal group and the control group when reading the corpus to further explore the process of consciousness processing.Part III: A study on the construction and predictive analysis of a model for identifying suicidal-tendency soldiers with metaphorical corpus.This part of the study analyzed the eye-movement features exhibited when reading suicidal tendency corpus(including metaphorical corpus and high surface validity corpus),extracted 11 eye-movement indicators that have psychological significance on reading studies,constructed recognition models by three machine learning algorithms,namely,plain Bayesian,supported vector machine and logistic regression,and analyzed the predictive validity of the models.Results:1.A qualitative study of the metaphorical soldier suicidal corpus found that suicidal visitors often used metaphorical language to express their internal emotions during interviews,and that their descriptions shared some degree of similarity,with metaphorical descriptions of psychological distress predominating.2.In the eye-movement study,several eye-movement indicators(e.g.sum of gaze duration,total number of gaze,gaze back time,mean pupil size)revealed that both groups of subjects processed the metaphorical corpus at a greater depth than the high surface validity corpus.3.Using the eye-movement data in Part II as the basis,the metaphorical corpus was classified using three machine learning algorithms,logistic regression,plain Bayesian,and support vector machine,with 85%,79%,and 82% accuracy in identifying suicidal people for the metaphorical corpus;and 77%,68%,and 77% accuracy in identifying suicidal people for the high surface validity corpus,respectively.All three classifiers showed higher accuracy rates for the metaphorical corpus than the high surface validity corpus.4.The model performance of the three models for the two corpora of suicidal tendency readers was evaluated.The AUC values of the three models were 0.94,0.89 and0.91 for logistic regression,plain Bayesian and support vector machine based on the metaphorical corpus,and 0.83,0.74 and 0.79 for the three models based on the high surface validity corpus,respectively.Conclusions:1.Both Part I and Part II of the study revealed significant practical implications of metaphorical language forms in suicide prevention work,providing new research perspectives on suicide prevention and intervention work.2.The selected eye-movement indicators both showed that the suicidal-prone group showed deeper cognitive processing on the metaphorical corpus compared to the control group;the suicidal-prone group showed significantly higher arousal on the metaphorical corpus than on the high surface validity corpus.3.The machine learning results showed that both the metaphorical corpus and the high surface validity corpus were better at identifying suicidal individuals by eye movement indicators,with the metaphorical corpus having a relatively higher recognition rate and better model performance for suicidal individuals. |
PDF Full Text Request