Modelling the impact of environmental and social determinants on mental health using generative agents.

Mental health is shaped by socio-environmental determinants, yet traditional research approaches struggle to capture their complex interactions. This review explores the potential of generative agents, powered by large language models, to simulate human-like behaviour in virtual environments for mental health research. We outline potential applications including the modelling of adverse life events, urbanicity, climate change, discuss potential challenges and describe how generative agents could transform mental health research.