Ethical considerations in the clinical application of prediction tools for severe mental disorders: perspectives from child and adolescent psychiatrists

The ability to predict the risk of severe mental disorders holds considerable promise for individuals at risk, potentially enabling prevention and early intervention. Although the clinical application of such predictive models in child and adolescent psychiatry remains a future prospect, it is essential to consider their social and ethical implications that their use may entail. This study explores child and adolescent psychiatrists' views on these issues through a cross-sectional online survey distributed to members of the European Society for Child and Adolescent Psychiatry. Of the 81 respondents, the majority identified the most significant benefits of using prediction tools as enabling earlier intervention by healthcare professionals (81.5%), improving the quality of care (77.8%), and helping families enhance their resilience (63%). Participants also expressed concern about potential harms, particularly violations of privacy (74.1%), discrimination (92.6%), and the lack of explainability in Artificial Intelligence algorithms (74.1%). While most participants recognise potential medical benefits in the clinical use of predictive tools, numerous concerns must be addressed before such technologies can be considered viable. These include unresolved ethical challenges, such as risks related to privacy, potential of stigmatisation and discrimination, and algorithmic opacity, as well as limitations of healthcare systems.