AI System in China Beats Human Doctors in Diagnosing Neurological Diseases


A hospital in China. / Photo by: Fcuk1203 via Wikimedia Commons


A Chinese AI system managed to beat top doctors in the country in diagnosing brain tumors and predicting hematoma expansion cases. The winner, BioMind AI system, has been developed by the Artificial Intelligence Research Center for Neurological Disorders at Beijing Tiantan Hospital.

In the study to test the AI for medical diagnosis, about 225 cases were used and two groups of doctors were included as comparisons. A team of researchers from the Capital Medical University successfully diagnosed 87 percent of the cases within 15 minutes using the AI. Meanwhile, the team composed of 15 senior doctors was only able to correctly diagnose 66 percent using the traditional methods.

Additionally, the AI system was able to accurately predict 83 percent of all brain hematoma expansion cases. But the group of doctors only managed to predict 63 percent of the cases. Brain hematoma is a condition wherein blood collects outside the brain. Most injuries to the brain and brain tumors can cause the condition.

The team behind BioMind AI used tens of thousands of images of the nervous system that were associated with diseases in order to train the system. The stock images have been obtained from the 10-year archive of Tiantan Hospital. Using the images, the AI can diagnose other neurological diseases including glioma and meningioma with an accuracy rate of at least 90 percent, similar to the potential of a senior doctor.

"I hope through this competition, doctors can experience the power of artificial intelligence. This is especially so for some doctors who are skeptical about artificial intelligence. I hope they can further understand AI and eliminate their fears toward it," explained Wang Yongjun, executive vice president at Tiantan Hospital.

As a clarification, Gao Peiyi, head of the department of radiology at Tiantan Hospital said that the human doctors performed above average accuracy as per typical hospital standards.