Researchers from Cambridge suggest using ChatGPT 4 to prioritize eye issues

Researchers from Cambridge suggest using ChatGPT 4 to prioritize eye issues

Artificial intelligence (AI) may help prioritize patients with vision problems, according to a university researcher.

According to Dr. Arun Thirunavukarasu, a University of Cambridge student, AI may determine whether situations are urgent enough to need a specialist's attention.

Researchers from Cambridge University put ChatGPT 4 to the test against the expertise of physicians at various career phases.

The AI algorithm performed well enough to get a 69% on an 87-question mock exam.

Training data for ChatGPT 4 included around 374 ophthalmic questions.

Three trainee ophthalmologists, two junior physicians without specialization, and five experienced ophthalmologists—doctor who treats patients with eye conditions—were compared to its responses.

They were also contrasted with various language models and a previous iteration of ChatGPT.

Compared to comparable apps like ChatGPT 3.5, Llama, and Palm2, ChatGPT 4 had a better score of 69%.

The median score for experienced ophthalmologists was 76%, for trainees it was 59%, and for younger physicians it was 43%.

Currently employed at Oxford University Hospitals NHS Foundation Trust, Dr Thirunavukarasu oversaw the research and completed the work as a student at the University of Cambridge's School of Clinical Medicine.

Speaking on the practical use of AI in patient triaging, he stated: "We can use it to determine which instances are urgent enough to visit a specialist right away, which can wait, and which don't need treatment.

"The models could follow clear algorithms previously in use, and we've found that ChatGPT 4 is as good as expert clinicians at assessing eye symptoms and signs to answer more complicated questions."Large language models may eventually be able to counsel general practitioners (GPs) who are having trouble getting timely guidance from eye physicians. More people than ever in the UK are waiting to get eye treatment."

Man had nine hospital appointments before receiving an eye cancer diagnosis.

A regular exam rescued the woman's vision.

ChatGPT now has access to current data.

Researchers suggested using the technology to triage patients, figuring out who needed specialized treatment and who can wait to visit a general practitioner.

Furthermore, they said that although language models like ChatGPT "do not appear capable" of taking the place of eye physicians, they might "provide useful advice and assistance to non-specialists".

In addition, Dr. Thirunavukarasu said that he thought physicians will still be in control of patient care.