Diagnostic reasoning prompts reveal the potential for large language model interpretability in medicine | Synapse