Competition will pit doctors against AI in identifying brain disease

World
Europe

Artificial intelligence (AI) has seized the public imagination in recent years, as highly publicized events like world-class Go players being outmaneuvered by computer program AlphaGo redefine what we think machines are capable of.

Meanwhile, governments are pouring money into the field to gain a leg up in sectors where humans have long reigned supreme, like transportation or healthcare. China even set the goal of becoming a global AI leader by 2030 in a State Council guideline issued last year.

Now the country’s foremost research center for neurological diseases will launch a competition in late June to pit doctors against AI in identifying some of the brain’s most serious problems. Contestants will diagnose patients using MRI and CT scans, a complex task that involves acute pattern recognition abilities.

The organizers say that it will train a proprietary program called “BioMind” with a database of cases from Beijing Tiantan Hospital, which houses the research center.

The hospital’s vice president and neurology department’s chief physician Wang Yongjun believes that smart diagnostic tools can eventually be used in clinical settings nationwide to provide high-quality service to patients far from centers of care, especially in small towns or rural areas.

Having machines help diagnose patients quickly and accurately could alleviate China’s overburdened healthcare system. A trained AI could reach a diagnostic accuracy rivaling that of an experienced physician, noted Wang at the press conference announcing the competition.

Will machines cast doctors aside?

Machine algorithms are already helping diagnose lung cancer and bone fractures that a human could miss.

A Stanford study found that the researchers’ machine learning model performed better than radiologists in detecting problems with fingers and wrists. However, the model did worse on elbows, forearms, and many other body parts.

The study goes on to conclude that the model could help prioritize images that are considered abnormal so that doctors can attend to patients with urgent need sooner, while also warding off “radiologist fatigue” from overwork, which can affect diagnostic accuracy.

Healthcare experts assert that such technology is primarily intended to assist medical professionals. In a journal article reviewing the current state of AI in healthcare, Wang and other authors note that though AI can help doctors make better clinical decisions, “human physicians will not be replaced by machines in the foreseeable future.”