Google's DeepMind Says Its AI Tech Can Spot Acute Kidney Disease 48 Hours Before Doctors Spot It (cnbc.com) 25
Five years after Google acquired DeepMind, the health and artificial intelligence group is unveiling its biggest breakthrough yet in health care. Its technology is able to predict if a patient has potentially fatal kidney injuries 48 hours before many symptoms can be recognized by doctors. From a report: In a paper published this week in the journal Nature, DeepMind researchers said their algorithms correctly predicted 90 percent of acute kidney injuries that would end up requiring dialysis. The work was the result of a project with the U.S. Department of Veteran Affairs to help doctors get a head start on treatment. "We've been really excited for the potential of using AI to support clinicians moving care from reactive to proactive and preventative," said Dominic King, DeepMind's co-founder and clinical lead, in an interview. About 2 million people die every year across the globe from acute kidney injury, according to researchers from the University of Pittsburgh School of Medicine. The condition, which involves a sudden episode of kidney failure or damage, can be tricky for doctors to diagnose because there aren't always immediate and clear symptoms. Studies have shown that catching it early can decrease the likelihood of serious injury or death.
Wow! (Score:1)
How desperate are these "AI breakthroughs" getting? Yeah, we get it: you can do image recognition in narrow controlled conditions (but not very well).
Re: (Score:2)
Re: (Score:2)
They can bump the "How to do surgery with a sawszall" video.
And yet (Score:2)
Re: (Score:2)
Will do.
Curious about the false positive rate (Score:2, Insightful)
I mean, I can achieve 100% accuracy in predicting all such kidney injuries if I always predict kidney failure on every. I'll never miss one. But that doesn't make it useful.
Re: (Score:3)
Scarece details (Score:2)
Re: (Score:2)
"Each clinical feature was mapped onto a corresponding high-level concept, such as procedure, diagnosis, prescription, laboratory test, vital sign, admission, transfer and so on. In total, 29 such high-level concepts were present in the data."
Not a really great explanation, but I don't really expect any better from papers in medicine. It's barely enough to reproduce, but not without a lot of effort. So for this paper to actually help anyone, they will need to either license the code, or someone else will have to do the work.
There are companies that are actually trying to apply this kind of thing practically in hospitals. The best success stories I've
Stupid net neutrality rollback (Score:2)
Next breakthrough, AI predicts how farts smell. (Score:2)
The article: https://www.cnbc.com/2019/07/3... [cnbc.com]
I read the article page 1, then Nature blurred the rest, but here is what page 1 says, boiled down:
"Our model predicts 55.8% of all inpatient episodes of acute kidney injury, and 90.2% of all acute kidney injuries that required subsequent administration of dialysis, with a lead time of up to 48h and a ratio of 2 false alerts for every true alert."
So, basically:
55% - They predicted half of the events. Abe Lincoln or Tom Jefferson in my pocket could have done th
Normal (Score:2)
The AI spots it Saturday morning while the doctors are doing 'weekend' until Monday morning, there's your 48 hours.