With 1.7 million deaths per year, lung cancer is the sixth leading global cause of death — deadlier than breast, prostate and colorectal cancers combined. Now Google wants to dramatically slash those terrible statistics with a new artificial intelligence-powered diagnosis system.
Detecting lung cancer early marks all the difference between a successful intervention and being too late to save a patient. Google claims that this AI-driven lung cancer detection can make this early detection possible with more accuracy, at a lower cost and at a larger scale than specialists can.
While Google stresses that its system is not a substitute for specialist (yet), it equally stresses the need for an automated early diagnostic tool: only 2 to 4 percent of eligible patients are screened today in the United States. If that’s appalling, imagine the situation in less developed countries across the world.
Published in the scientific journal Nature Medicine, the paper says that the new AI program can match or outperform the diagnosis of six expert radiologists over 45,856 single computer tomography scans from the National Institute of Health’s dataset compiled by the National Lung Screening Trial study and Northwestern University.
“We detected five percent more cancer cases while reducing false-positive exams by more than 11 percent compared to unassisted radiologists in our study,” the paper says, an accuracy rate of 94.4 percent.
For all the dystopian doom and gloom about AI exterminating all of humanity, it seems that the technology can bring a dramatic change to the way we diagnose and heal people across the entire planet without requiring them pay through the nose for treatment.