You are using an older browser version. Please use a supported version for the best MSN experience. launches artificially intelligent medical software to find abnormalities faster

TechCrunch TechCrunch 9/05/2016 Sarah Buhr

Jeet Raut’s mom was told she no longer had breast cancer. But it turned out to be a false diagnosis and she had to undergo further treatment.

She’s okay now, but that medical mistake could have cost her life and it gave Raut the idea to build a better way to catch medical abnormalities in the body.

He and his co-founder Peter Wakahiu Njenga created to speed up the process of finding cancers and minimize human error.

“The idea behind is to increase efficiency,” Raut told TechCrunch.

A new report from BMJ puts medical error as the third leading cause of death in the United States. And radiologists must comb through an increasing number of body scans every year – the number has tripled to 149 CT scans per 1,000 patients since 2012.

“Doctors now have more data than ever, but if radiologists had to read every single image they’d never get through them,” Raut said of the problem.

Thanks to advances in machine learning, Raut and Njenga figured they could teach a program to do the same thing, only better and faster.

They’re an impressive duo – both studied at Columbia and then Njenga went on to UC Berkeley and later worked on Facebook’s machine learning team. Raut continued to the University of Illinois at Urbana-Champaign and later Stanford for life extension research the Computers and Cognition Lab. is an outgrowth of both Njenga and Raut’s training. It works by feeding the software algorithm hundreds of scans of healthy vs. unhealthy lungs, for example, to teach the program how to identify a problem and then improves on its own over time.

How much better is it than a human? Raut admits the accuracy of the algorithm isn’t 100 percent. He wasn’t sure the number but guessed it was closer to 85 percent.

“We’re initially focused on increasing the efficiency of the doctor while helping them maintain accuracy, but down the line we’d like to help them become more accurate,” he said.

There’s also the trust factor. Hospitals are notorious for bureaucracy and slow to adopt new technologies. Raut brushes off the suggestion hospitals may not want what he’s got and tells me is in talks with several large hospitals for possible partnerships in the near future. will still face stiff competition from IBM Watson and others in the AI space getting in on the medical market and it will need to pass FDA regulations (which no one has been able to do so far, but IBM has been pushing Congress on).

More from TechCrunch

image beaconimage beaconimage beacon