Integrating Digital and AI Technologies into Mental Health: Lessons from a Four-Year SG-NZ Collaboration on Multi-Modal AI Models.

The LYRIKS dataset is a unique study on Singaporean youths at ultra-high risk of developing a spectrum of mental health disorders. Although the original study concluded over 13 years ago, we continue to uncover novel findings. This ongoing discovery is fuelled by the availability of new digitization modalities, including omics and natural language processing, as well as advanced AI technologies.

Through a four-year project funded by Singapore and New Zealand, my team explored the potential of novel bio and digital markers, ranging from genes to text. We also developed various new feature selection techniques and innovative network-based models. I will present a selection of notable works from this collaboration.

Implementing AI and digital technology in healthcare goes beyond creating models that excel in academic publications. Stakeholder engagement is critical. Clinicians, although receptive, have reservations. Developers, who play a crucial role in building AI models, seek greater involvement. It is also essential to address and include the perspectives of patients, the public, and minority groups. I will discuss some of our ongoing efforts to incorporate stakeholder perspectives to enable successful digital transformation.

Finally, I would like to share some thoughts for future research. Specifically, using context to enhance data integration, how to connect digital phenotyping with earlier findings to enhance mental health monitoring and management.