Slashdot Mirror
Google Is Training Machines To Predict When a Patient Will Die (bloomberg.com)
A newly developed tool by Google can forecast a host of patient outcomes, including how long people may stay in hospitals, their odds of re-admission and chances they will soon die. Google documented some of this tool's abilities in May; in one instance, Google's tool estimated, by taking 175,639 data points into consideration, that a particular patient's odds at dying during her stay at the hospital was 19.9 percent, up from 9.3 percent that the hospital's computers had estimated. Now Bloomberg reports what Google intends to do with this new tool next. From the report: Google's next step is moving this predictive system into clinics, AI chief Jeff Dean told Bloomberg News in May. Dean's health research unit -- sometimes referred to as Medical Brain -- is working on a slew of AI tools that can predict symptoms and disease with a level of accuracy that is being met with hope as well as alarm. Inside the company, there's a lot of excitement about the initiative.
"They've finally found a new application for AI that has commercial promise," one Googler says. Since Alphabet's Google declared itself an "AI-first" company in 2016, much of its work in this area has gone to improve existing internet services. The advances coming from the Medical Brain team give Google the chance to break into a brand new market -- something co-founders Larry Page and Sergey Brin have tried over and over again. Software in health care is largely coded by hand these days. In contrast, Google's approach, where machines learn to parse data on their own, "can just leapfrog everything else," said Vik Bajaj, a former executive at Verily, an Alphabet health-care arm, and managing director of investment firm Foresite Capital. "They understand what problems are worth solving," he said. "They've now done enough small experiments to know exactly what the fruitful directions are." The report adds that, among other things, Google's tool has the ability to sift through notes buried in PDFs or scribbled on old charts.
"They've finally found a new application for AI that has commercial promise," one Googler says. Since Alphabet's Google declared itself an "AI-first" company in 2016, much of its work in this area has gone to improve existing internet services. The advances coming from the Medical Brain team give Google the chance to break into a brand new market -- something co-founders Larry Page and Sergey Brin have tried over and over again. Software in health care is largely coded by hand these days. In contrast, Google's approach, where machines learn to parse data on their own, "can just leapfrog everything else," said Vik Bajaj, a former executive at Verily, an Alphabet health-care arm, and managing director of investment firm Foresite Capital. "They understand what problems are worth solving," he said. "They've now done enough small experiments to know exactly what the fruitful directions are." The report adds that, among other things, Google's tool has the ability to sift through notes buried in PDFs or scribbled on old charts.
Google fixes false positive rate of patient death predictor machines by training another machine to kill patients predicted to die.
The only thing necessary for evil to triumph is for it to be pitted against a slightly greater evil
Here's the problem: how does one avoid a bad feedback loop? i.e. Computer predicts patient is likely to die, so doctors spend less time attempting to work the problem and/or shift the patient into a palliative care pathway. By predicting that a patient is likely to die, the computer will have made that patient EVEN MORE likely to die.
"The Google AI reports of my death are greatly exaggerated."
It must have been something you assimilated. . . .
I have. It takes some real desperation to make that happen. The only way to make that worse is to seperate said 2 year old from it's parents and lock it away in what amounts to a refurbished dog pen as if the 2 year old is even capable of criminal culpability. Even Trump recognized that that was sufficiently repugnant that he'd better try to blame someone else for it.
Oddly, you simultaneously cheer for the action and Trump who now says the action is wrong.You also seem to believe what Trump says, but stand against the people Trump says are responsable for the action you applaud.
Orwell was right about Doublethink, but wrong about who would institute it.
This could be used for good purposes and bad. We know which most corporate financials would choose.
Greater than 50% chance of dying in hospital, time to become an outpatient. The good, perhaps your chances of dying reduce once you leave, hospital that is :)
And insurance policies that have a clause that states if your chance of dying while in hospital is greater than 50% then there's no cover for treatments only death. The good, perhaps the cost of the cover goes down.
When shit hits the fan get some of these https://youtu.be/pY-GncsZ-UE
Algorithm: At an arbitrary chance of dying, withhold or delay insurance approval of a select (expensive) set of treatments. Business Rationale: 90% of an average person's lifetime medical costs are in the last year of life. Of that figure, another 90% of the costs are in the last 3 months of life. Optimizing the cutoff for medical treatment will maximum profits.
My real question is which institution is in massive HIPAA violation with giving Google all this data?
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
It is possible and common to both use anonymized patient data and remain HIPAA compliant.
They decide which people get insurance, and how good the coverage is.
I'll see your senator, and I'll raise you two judges.
Our hospital has a betting pool that fairly accurately anticipates the time of death for a patient. Some of us are more accurate than others, of course, and make more money in the process. But it's all in good fun and we all get better in our predictions. For my part, I've won three times in three months, and overall that means I've won slightly more than I bet. If you're expecting to die, I hope you visit our hospital. You are welcome to bet along with us!
...omphaloskepsis often...