Exploring Mental Health Issues - A data Analytics Approach

Abstract

The emergence of COVID-19 and its associated containment strategies, such as lockdowns and social distancing, are expected to impact mental health, and this impact could be more severe among people with pre-existing mental health disorders. In this research project, we aim to understand better the changes in mental health in the time of COVID-19 pandemic by analysing data from mental health-related communities. We have collected data from the 15th of February 2020 to the 15th of July 2020 and analysed this data using interaction, linguistic structure and interpersonal awareness measures. Our findings have shown that early in the lockdown, individuals showed selflessness, solidarity and low rates of seeking help, but after that they showed a negative mental health state. Moreover, considering the importance of social support in mental illness, we also aim to explore what derives social support in mental health communities. We found that receiving high social support was hindered by the use of more swearing and negative emotional or self-referent words. Furthermore, not receiving social support may push actual help seekers to repeat their posts, which may be considered a spam action. Therefore, we aim to investigate what characterises duplicate posts authored by actual help seekers to build a classifier model that is able to classify them as not spam actions. Our investigation showed that actual help seekers tend to show different levels of mental health when they repeat their posts for seeking immediate help.