The Prediction Of Emotion And Stress Based On Personality Demographics And Event-Related Potentials

Our brain is sensitive to stimulus associated with negative emotion.Particularly,when perceiving negative and threatening stimulus,individuals tend to trigger stress responses.Overreacted stress responses may hamper cognitive function and lead to mood disorders such as depression and anxiety.Therefore,individualized ability prediction of emotion regulation and stress response with physiological and psychological characteristics is desirable,not only for protecting at risk individuals,but also for improving employment for government and enterprise on positions requiring high emotional stability.In this research direction,the present study examined predictive values of(1)resilience features for event-related potentials(ERP)measuring negative emotional response,and(2)support vector regression(SVR)for cortisol levels measuring stress response.For(1),we examined the linear relationship between resilience,as well as its composite factors,and the late positive potential(LPP)evoked by negative emotional response.When LPP measures individuals' ability in emotion regulation,our results showed a negative correlation between resilience and this ability indicator.Additionally,this negative correlation was found to be mainly driven by the composite factor of optimism.For(2),we fed 61 variables of demographic characteristics,personality,as well as task-based ERP into a SVR model for prediction of salivary cortisol level after a stress challenge.With 17 variables excluded with recursive feature elimination,the remaining 44 features maximized the prediction accuracy with the goodness of fit(R^2)between predicted and true value reached 0.753.This SVR model enormously outperformed all the general linear approaches we previously used in cortisol level prediction.The present results indicate that resilience factors can be used in predictive models for negative emotional response.Additionally,with its superior performance,support vector regression is expected to be optimized and increasingly utilized in future applications of emotion and stress prediction.