Key points are not available for this paper at this time.
Mental health is a critical issue worldwide and effective treatments are available. However, incidence of social stigma prevents many from seeking the support they need. Given the rapid developments in the field of large-language models, this study explores the potential of chatbots to support people experiencing depression and anxiety. The focus of this research is on the engineering aspect of building chatbots, and through topology optimisation find an effective hyperparameter set that can predict tokens with 88.65% accuracy and with a performance of 96.49% and 97.88% regarding the correct token appearing in the top 5 and 10 predictions. Examples of how optimised chatbots can effectively answer questions surrounding mental health are provided, generalising information from verified online sources. The results of this study demonstrate the potential of chatbots to provide accessible and anonymous support to individuals who may otherwise be deterred by the stigma associated with seeking help for mental health issues. However, the limitations and challenges of using chatbots for mental health support must also be acknowledged, and future work is suggested to fully understand the potential and limitations of chatbots and to ensure that they are developed and deployed ethically and responsibly.
Bird et al. (Wed,) studied this question.