This one day workshop is organised by Angus Roberts and Rob Stewart from DataMind - The Health Data Research Hub for Mental Health, the hub for mental health informatics research development. Whilst DataMind has a specific focus on mental health, it is expected that the workshop will be of interest and relevance to people from across the entire health text analytics community. The workshop will consider application of the "New NLP" to all aspects of mental health: research, clinical care and therapy.

The advent of large language models and generative AI is bringing about a sea change in natural language processing, has provided new perspectives on many of the challenges that have faced health-related NLP, and is opening up new research directions. We can now extend the reach of NLP, with the processing of complex language that assumes domain expertise and pragmatic understanding becoming more tractable. We have new solutions to perennial problems: where once we talked about the challenges of obtaining access to restricted health record text, we can now realistically discuss generating high-quality synthetic records, and where we struggled to find domain expert time for data labelling, we now consider using generative AI annotators. Beyond research, many are considering ways in which generative AI can play a role in therapeutic settings. The new NLP also raises many concerns, including those of bias; the ethics and ownership of models built from patient data; and the availability, financial and environmental cost of the processing power required to train ever-larger models.

This one day workshop will bring together attendees from informatics, clinical, industry and service user backgrounds to discuss the shift in NLP and what it means for mental health. The morning will focus on technical challenges and innovations, with the afternoon given to papers and discussions of broader operational, clinical, and societal implications.