ChatGPT-4 Outshines Humans in Medical Probability Assessment
In a groundbreaking development, ChatGPT-4, the latest advancement from OpenAI, has demonstrated its superiority over human clinicians in the nuanced realm of probabilistic reasoning. The research, detailed in a research letter published on December 11 in JAMA Network Open, reveals the AI’s remarkable proficiency in determining pretest and post-test disease probability, particularly after a negative test result involving chest radiographs and mammograms.
Led by Dr. Adam Rodman from Beth Israel Deaconess Medical Center in Boston, the investigative team found that ChatGPT-4 excelled in minimizing errors related to pretest and post-test probability after a negative outcome in five distinct cases. These cases involved chest radiography for pneumonia, mammography for breast cancer, stress tests for coronary artery disease, urine culture for urinary tract