The study explores the use of large language models (LLMs) to support clinical decision-making in bipolar depression. The researchers developed 50 clinical vignettes reflecting bipolar depression and presented them to experts in bipolar disorder, who were asked to identify optimal and poor or contraindicated pharmacotherapies. The same vignettes were then presented to an LLM (GPT4-turbo) with or without prompts based on recent bipolar treatment guidelines. The augmented model prioritized the expert-designated optimal choice in 508 out of 1000 vignettes (50.8%), compared to 234 out of 1000 for an un-augmented model (23.0%). Community clinicians, on average, identified the optimal choice in 23.1% of vignettes. The augmented model showed promise as a scalable strategy for clinical decision support, but further studies are needed to address potential biases and overreliance on such models. The study highlights the potential of integrating treatment guidelines into LLMs to improve clinical outcomes in bipolar depression.The study explores the use of large language models (LLMs) to support clinical decision-making in bipolar depression. The researchers developed 50 clinical vignettes reflecting bipolar depression and presented them to experts in bipolar disorder, who were asked to identify optimal and poor or contraindicated pharmacotherapies. The same vignettes were then presented to an LLM (GPT4-turbo) with or without prompts based on recent bipolar treatment guidelines. The augmented model prioritized the expert-designated optimal choice in 508 out of 1000 vignettes (50.8%), compared to 234 out of 1000 for an un-augmented model (23.0%). Community clinicians, on average, identified the optimal choice in 23.1% of vignettes. The augmented model showed promise as a scalable strategy for clinical decision support, but further studies are needed to address potential biases and overreliance on such models. The study highlights the potential of integrating treatment guidelines into LLMs to improve clinical outcomes in bipolar depression.