Progress and challenges in the analysis of big data in social media of adolescents

Social media are ubiquitous today, and adolescents use them to express their thoughts, feelings, and behaviours. New interdisciplinary methods allow the automatic analysis of the massive amounts of data (big data) available on social networking websites using machine-learning tools to detect indicators of mental-health problems and disorders by identifying differences with common activity and communication patterns. This review first introduces the concept and potential fields of applications of big data in social media. It then discusses the first studies that used big data analyses and detected mental-health problems by identifying differences in the structure of social networks, in the use of certain words, and in the communication of opinions and sentiments. Future studies employing several assessment points could use longitudinal mediation analysis to model intraindividual changes in order to understand when and through which mechanisms social media use has an impact on mental health. Furthermore, future studies should include additional mental disorders, various sources of information, a broader age range, and additional social-networking websites to develop more precise models for the early detection of mental disorders. This would enable the development of personalised intervention programs to promote mental health and resilience in adolescents.