Anticipating the future promise of AI in medicine


https://www.statnews.com/2019/02/14/artificial-intelligence-medicine-eric-topol/?utm_source=The+Weekly+Gist&utm_campaign=41103e2ef1-EMAIL_CAMPAIGN_2019_02_14_09_16&utm_medium=email&utm_term=0_edba0bcee7-41103e2ef1-41271793

Image result for artificial intelligence in healthcare

A group of American and Chinese researchers published data this week showing that artificial intelligence (AI) is as accurate as physicians in diagnosing common clinical conditions in children. Scientists built an AI model using neural networks to process patient history, physical exam and lab data, clinical symptoms and other information to automatically generate a diagnosis. Using that model to evaluate the records of over 600,000 Chinese pediatric patients, the diagnostic accuracy of the AI-driven model was largely equivalent to that of physicians. Looser privacy standards in China make it easier to aggregate the data for AI-driven diagnosis, presenting a potential roadblock for replicating the results in the US. However, researchers cite the potential for AI to complement physician diagnosis, as algorithms recognize patterns that are often missed by doctors.

The scale of this study is impressive, but it’s hardly the first to illustrate the promise of AI in improving diagnosis and even substituting for high-cost clinical labor. However, few AI technologies have been able to make the leap from promising algorithm to real clinical application. Writing in Nature Medicine, digital-medicine guru Dr. Eric Topol recently reviewed the science and application of AI across clinical care, and found that while he “couldn’t find one discipline in medicine that doesn’t have significant AI potential impact”, there is an “AI chasm” between the developing science and real clinical impact. Most AI research is retrospective, and Topol identifies the need for true gold-standard, prospective studies. But he says that real impact, likely in visual diagnosis, could be imminent, with studies demonstrating AI analysis of radiographic images, retinal scans and skin lesions that is equal to or better than a doctor’s read. Topol doesn’t cite one key barrier of AI implementation: professional guilds, who have vested interest in keeping the diagnostic business in the hands of their members. Regardless, AI represents a promising path to reducing reliance on expensive human labor, one that is sure to be adopted as cost pressures mount. While we’d predict the first impact will come from automating “back-office” functions, doctors who resist AI are fighting a losing battle. 

Successful physicians will ascertain how to use AI to augment their practice—and the ones who blindly resist its use may be most in danger of being rendered obsolete.

 

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.