A new study suggests that a generative artificial intelligence program called GPT-4 may be more accurate than clinicians in diagnosing elderly patients with complex medical histories and long hospital stays.
The study found that GPT-4 accurately diagnosed four out of six patients, while clinicians only accurately diagnosed two out of six patients! When differential diagnoses were included, GPT-4’s accuracy improved to five out of six patient diagnoses, compared to three out of six for clinicians. It seems that GPT-4 could potentially help clinicians analyze clinical situations with diagnostic difficulties, such as identifying underlying malignancies or drug side effects.
However, GPT-4 had difficulty with certain aspects of diagnosis, such as pinpointing the source of recurrent infections. The study authors noted that GPT-4 should be seen as a tool to increase a clinician’s confidence in a diagnosis or offer suggestions similar to those of a specialist, especially in lower-income countries with limited access to specialists. The study had a small sample size and the use of GPT-4 assumes accurate medical history information. Despite the study’s challenges, the use of generative AI in diagnosing patients with extensive medical histories shows great promise.
The whytry.ai article you just read is a brief synopsis; the original article can be found here: Read the Full Article…
