Beyond the prompt: gender ratio in text-to-image models, with a case study on hospital professions | Synapse