Google has trained its AI to help identify lung cancer

According to the World Health Organization, more than 1.7 million people die from lung cancer around the world every day.

Overall, the disease is the sixth most common cause of death worldwide.

For that reason, Google has been working since late 2017 on how to use artificial intelligence to help doctors identify lung cancer.

As Google notes, doctors typically have to examine hundreds of images after performing a CT scan on a patient. All the while, traces of cancer can be difficult to spot.

To help with this, Google has trained its AI to predict general cancer malignancy and identify subtle malignant tissue in the lungs. Further, Google’s AI model can even factor in information previous scans to compare the growth rate of suspicious lung nodules.

Google says it tested this model using nearly 46,000 chest CT screens, which it compared to the results of six board-certified radiologists. According to Google, its AI detected five percent more cancer than the radiologists and reduced false-positive exams by more than 11 percent.

That said, Google noted it still has work to do before the AI can be deployed in actual hospitals.

Going forward, Google says it’s working with its Cloud Healthcare and Life Sciences divisions to improve the AI model. Additionally, the company says it’s in “early conversations” with partners around the world for further clinical research and deployment.

Google asks anyone from a research institution or hospital system who’s interested in collaborating future AR research by filling out this form.

Source: Google