A Systematic Review of Artificial Intelligence and Mental Health in the Context of Social Media

Objective: We conducted a systematic review on the use of Artificial Intelligence (AI) in the psychology domain within the context of social media's role on mental health. We identified the types of mental health studies that AI has supported, reviewed the machine learning methods in these studies, and reported on the different approaches for data collection. We provided a critical review of the applicability of these methods in real-world settings. Finally, we discussed the challenges faced in this area of study and provided advice for other researchers interested in solving these issues in the future. Methods: We collected our data from three outlets: ACM, JMIR, and CLPsych. We focused on the studies on the application of AI in clinical psychology in the past six years (2016 to 2021). Results: A total of 37 articles were included in our study for further review. The number of publications increased over time. While CLPsych has the highest number of articles, Reddit was the commonly used social media site for data collection. Suicide was the most mentioned mental disorder mentioned in the studies. SVM was the most frequently used approach when applying AI in mental health studies. Conclusion: Our review of the existing literature identified three issues on the topic. First, social media is underutilized for mental health care. Second, there is a lack of collaboration between seemingly disconnected research communities: i.e., machine learning experts and clinicians. Third, little attention is paid to humans when conducting the research.