Sunday , June 20 2021

The Chinese study shows the potential of AI in the medical diagnosis – quartz



A new study from China found that an artificial intelligence system could best cure some doctors when it comes to diagnosing common childhood illnesses. The study, published in Nature Medicine yesterday (February 11th), is a co-author of a long list of Chinese and American researchers, and is just one of the latest in China to explore the use of artificial intelligence in medical diagnostics.

The study prepared a 101 million data point training system generated by electronic records of 1.3 million patient visits at a medical center in Guangzhou. Researchers found that the artificial intelligence system was able to meet or surpass two groups of junior doctors in accurately diagnosing a range of diseases, from asthma and pneumonia, from sinusitis and mouth-related illness. AI is also able to meet or exceed diagnostic parameters with some groups of senior doctors, for example, in the category of upper respiratory problems.

In some cases, the system is able to diagnose conditions with an accuracy of 90 to 95%. In no AI category, the model is not lower than the diagnostic accuracy level of about 79%, higher than a group of junior doctors, but lower than the other. Senior doctors generally did better than artificial intelligence.

One limitation on the study, however, is that the researchers downloaded all their data from visits to a medical facility in China, raising questions about its applicability abroad, and even the facility itself in Guangzhou. (As is often the case with AI systems, the algorithm tests the same data sources on which the model is based, raising additional questions about reproducibility.)

Still, the study adds to the growing research demonstrating the benefits of AI when it comes to diagnosis. For example, a study published last October also from China shows that another algorithm that predicts exactly 88% of the time whether the patients in a vegetative state will wake up. This happened a few months after the US Food and Drug Administration approved an AI system to diagnose diabetic retinopathy, which could lead to blindness if not treated.

Another work has also shown the usefulness of AI in imaging-based diagnostics – last year an artificial intelligence system defeated physicians in accurately diagnosing brain tumors in a Beijing competition. However, it is important to note that these systems are more likely to serve as a useful diagnostic tool for doctors than to replace them quite soon.

It is hardly surprising that much of this research comes from China. The country has invested heavily in artificial intelligence research, with the number of government-related documents on the topic rising by 400% since 2007. and urban areas, to cope with the aging of the population and to improve the efficiency of the health care system that many doctors and nurses need.


Source link