An international team of scientists has shown that an artificial intelligence system designed to diagnose childhood illnesses can recognize the symptoms more accurately than many doctors, Nature magazine reported today.
Designers programmed specific words and phrases in the system, in addition to digital data such as the patient's temperature typical of various diseases. About 101.6 million data were used to train the program.
The results show that artificial intelligence is able to diagnose glandular fever (also known as mononucleosis), pink, flu, chickenpox and human foot-and-mouth disease with an accuracy of between 90 and 97 percent.
Researchers acknowledge that the accuracy of AI diagnostics is better than that of physicians with little experience, although most experts perform the task with greater success than the program.
London, February 13, 2019
Chronic Digital / PL