Google (NSDQ:GOOG) has unveiled a new artificial intelligence system that outperformed six radiologists in detecting lung cancer, identifying 5% more cancers and cutting false positive rates by 11%, according to a STAT News report.
The system was trained on 42,000 computed tomography scans taken in a National Institutes of Health clinical trial, according to the report.
Google’s AI-powered system, which the company said is not intended to replace radiologists, also performed on par with radiologists when evaluating cases that included prior patient images, STAT reports.
Lung cancer kills more Americans than any other cancers, according to the report, with approximately 160,000 deaths reported last year.
Data from the NIH-study Google used to train its AI showed that screening of high-risk patients can reduce the risk of death by approximately 20%, though it raised concerns over the harm that can be caused by false positives, according to the STAT report.
“These people have a technology that will improve the precision of screening tremendously. It’s going to prevent more bad things from happening to people who are being screened,” Dr. Otis Brawley of John Hopkins University told STAT.
The system will still require more rigorous testing, including a randomized controlled trial, before it would see use in an actual medical practice, but the company said they are working with clinical partners to refine and validate the system.