Share this story!
- ChatGPT was compared to doctors in a Dutch study.
- Version 4 of ChatGPT made the correct diagnosis in 97 percent of cases.
- Researchers see potential for artificial intelligence in emergency departments.
Same results as doctors
According to a new study conducted in the Netherlands, it appears that ChatGPT might be as effective as doctors when it comes to diagnosing patients who visit the emergency department, reports Swedish news agency TT.
In the study, the researchers input doctors' notes about the symptoms of 30 patients and results from physical examinations into two versions of ChatGPT. Afterwards, they evaluated how the system's diagnostic suggestions aligned with the doctors' and the final diagnosis.
When the results were compared, doctors had the correct diagnosis in their top five suggestions in 87 percent of the cases. The same numbers were recorded for version 3.5 of ChatGPT. However, the later version 4.0 of ChatGPT got the correct diagnosis in 97 percent of the cases.
"There is potential to use this in the future to save time and reduce waiting times in the emergency room. Artificial intelligence could also be a tool for less experienced doctors or for detecting rare diseases," says Hidde ten Berg at Jeroen Bosch Hospital, one of the researchers behind the study.
The study was published in the scientific journal Annals of Emergency Medicine.