Impact of the large language model and prompt specificity in generating psychiatric multiple-choice questions | Synapse