"Clinical Reach into the Cognitive Space" (CRITiCS) - An outline conceptual framework for the safe use of generative artificial intelligence in mental health decision making.

Abstract

This article provides a conceptual framework - “CRITiCS” - for potential use in the design and deployment of generative AI systems to support clinical decision making in mental health services. Recent advances in generative AI, via massive parameter transformer networks (large language models such as GPT4) have rendered possible the use of cognitive agents that may support clinical productivity in mental healthcare, particularly in secondary and tertiary services where information load is proportional to clinical complexity and chronicity. Thus, the way we conceptualise the ‘cognitive space’ that, up to 2023, was occupied solely by human clinical reasoning, will inevitably change. However, there are transparency issues in that the cognitive emulation observed in LLM outputs, while of obvious potential benefit as an analogue of clinical reasoning,is not amenable to granular understanding in the same way that clinical decision making is subject to interrogation by human governance and assurance bodies. A consensus conceptual framework is urgently needed, to direct the path of the use of generative agents in mental health care, which we argue has cognitive space features that distinguish it from physical medicine, towards which most AI deployment proposals to date have been directed. It is hoped that this paper will attract discussion and interest in both the clinical and development communities.

Formulation as Representation - Modelling the Cognitive Space of Mental Health Clinical Reasoning

“…This is because the degree of cognitive and administrative work required by mental healthcare clinician, particularly when working with the most complex problems, is one of the highest in all areas of healthcare (Stade et al 2023). Previous papers have argued that the use of generative AI in mental health requires particular safety diligence due to the interpersonal nature of mental healthcare (Hider and Wright 2023). In this paper, an overview is given of the specific representational features of clinical case formulation as used in mental healthcare.…”