Information Extraction from Clinical Notes: Are We Ready to Switch to Large Language Models?