Researchers at James Cook University led by a data science lecturer, Usman Naseem, are exploring a novel approach to better support individuals facing mental health concerns by analyzing their social media posts, applying a new technique designed to identify early warnings of mental health distress in online expressions of people.
Dr Naseem termed the global emergence of mental health conditions a pressing concern, with more than one in 100 deaths attributed to suicide. He says the Covid-19 pandemic exacerbated this issue, leading to a notable increase in conditions such as depression and anxiety, affecting over one billion people worldwide.
To address this urgent need for innovative approaches to mental health concerns, Dr Naseem's team focused on user-generated text, particularly on social media platforms, as a valuable tool. Social media users often express themselves openly, providing a rich source of data for researchers.
By discerning changes in sentiment, identifying specific language markers, and detecting behavioural anomalies, researchers can potentially spot risk factors for mental health conditions, he believes.
The technique was refined by considering historical posts, their timing, and the intervals between them.
Dr Naseem says that accurately assessing a user's mental state requires an understanding of the history of their mental health condition. By comprehensively analyzing both historical posts and the diverse time intervals between them, the researchers aimed to provide more accurate and nuanced assessments of a person's mental well-being.
According to Dr Naseem, the new technique effectively captured the context of users' historical posts and irregularities in the timing of their posts. The results demonstrated that this method surpasses current state-of-the-art approaches in mental health surveillance on social media.
The researchers anticipate that this innovative approach will offer valuable insights for mental health professionals and researchers in monitoring and supporting individuals in need.
Ideally, the information derived from this research could become an efficient and effective way for clinicians to provide early intervention and support.