China might be gaining an edge in the artificial intelligence race — particularly in healthcare — on account of its less-stringent privacy regulations, according to an article in Wired.
Five notes:
1. In the U.S., companies often struggle to gather enough health data to train an accurate AI algorithm, in part because of strict privacy rules that require them to negotiate with multiple partners who have access to this data. IBM, for example, has spent more than $3.5 billion since 2015 acquiring software companies to amass data sets of patient records.
2. China's government has made healthcare AI a priority, as outlined in a national AI strategy the country launched in 2017. While China has been bolstering its protections on patient data — the country already mandates researchers use anonymized records — there are fewer rules and regulations, according to Wired.
Eric Topol, MD, director of the Scripps Translational Science Institute in La Jolla, Calif., told Wired: "China has a decided advantage with respect to the quantity of data and the nurturing of the government."
3. Infervision, a Beijing-based startup founded four years ago, has already amassed more than a million scans from Chinese hospitals to train and test its AI algorithms for radiology. The startup creates software to assist radiologists — such as to identify lung nodules — and its chief scientist agrees that its access to data gives it an advantage over U.S. companies.
"In the U.S., particularly for big academic hospitals, you have to go through so many processes and it can take a really long time to access data," the chief scientist, Yufeng Deng, PhD, told Wired. "In China it's less well-defined."
4. Infervision is seeking FDA approval to market its AI product in the U.S., and it already provides software to Raleigh, N.C.-based Wake Radiology and Palo Alto, Calif.-based Stanford Children's Health. These U.S. partners work with Infervision to test its software for free and provide feedback on any errors they find. To work with American patients, Infervision also refined its algorithm with U.S.-sourced data.
Wake Radiology CIO Matt Dewey told Wired his organization is testing algorithms from other startups, in addition to Infervision.
5. Dr. Topol highlighted that China's current edge in easing the process for acquiring data doesn't mean companies in the country will beat out ones in the U.S. As an example, he noted it's still unclear how well AI trained on Chinese patients and equipment will perform when asked to analyze patients and equipment in the U.S.
To read Wired's article, click here.