Hospitals May Turn To Algorithms To Fight Fatal Infections (scientificamerican.com) 55
An anonymous reader quotes a report from Scientific American: Clostridium difficile, a deadly bacterium spread by physical contact with objects or infected people, thrives in hospitals, causing 453,000 cases a year and 29,000 deaths in the United States, according to a 2015 study in the New England Journal of Medicine. Traditional methods such as monitoring hygiene and warning signs often fail to stop the disease. But what if it were possible to systematically target those most vulnerable to C-diff? Erica Shenoy, an infectious-disease specialist at Massachusetts General Hospital, and Jenna Wiens, a computer scientist and assistant professor of engineering at the University of Michigan, did just that when they created an algorithm to predict a patient's risk of developing a C-diff infection, or CDI. Using patients' vital signs and other health records, this method -- still in an experimental phase -- is something both researchers want to see integrated into hospital routines.
The CDI algorithm -- based on a form of artificial intelligence called machine learning -- is at the leading edge of a technological wave starting to hit the U.S. health care industry. After years of experimentation, machine learning's predictive powers are well-established, and it is poised to move from labs to broad real-world applications, said Zeeshan Syed, who directs Stanford University's Clinical Inference and Algorithms Program. Shenoy and Wiens' CDI algorithm analyzed a data set from 374,000 inpatient admissions to Massachusetts General Hospital and the University of Michigan Health System, seeking connections between cases of CDI and the circumstances behind them. The records contained over 4,000 distinct variables. As it repeatedly analyzes this data, the ML process extracts warning signs of disease that doctors may miss -- constellations of symptoms, circumstances and details of medical history most likely to result in infection at any point in the hospital stay.
The CDI algorithm -- based on a form of artificial intelligence called machine learning -- is at the leading edge of a technological wave starting to hit the U.S. health care industry. After years of experimentation, machine learning's predictive powers are well-established, and it is poised to move from labs to broad real-world applications, said Zeeshan Syed, who directs Stanford University's Clinical Inference and Algorithms Program. Shenoy and Wiens' CDI algorithm analyzed a data set from 374,000 inpatient admissions to Massachusetts General Hospital and the University of Michigan Health System, seeking connections between cases of CDI and the circumstances behind them. The records contained over 4,000 distinct variables. As it repeatedly analyzes this data, the ML process extracts warning signs of disease that doctors may miss -- constellations of symptoms, circumstances and details of medical history most likely to result in infection at any point in the hospital stay.
Hospitals May (Score:3)
Re: (Score:2)
*snort*
Re: (Score:3)
I was wondering the same thing, actually.
If the infection can be fought of, by definition it wasn't fatal.
I'm thinking that they probably meant to use the word "life threatening", or perhaps even "critical".
I suppose an infection could still be considered fatal if it kills you, but you are still successfully resuscitated afterwards. But that's really kind of reaching, if you ask me.
Re: (Score:2)
Isn't fighting a fatal infection somewhat of a waste of resources?
If the patient dies, but is infected with a zombie virus, you might want to continue treatment to prevent the corpse from turning into a zombie.
Either that or just flamethrower the bastard.
May turn to them? (Score:5, Informative)
We are already doing that. We use predictive analytics that processes dozens of data elements on each patient in the hospital and scores them for sepsis risk. The system then can do many things with that score. The most popular is paging to the attending provider and care team. This helps to reduce the cases of septic shock significantly.
Re: (Score:2)
It would be nice if Hospitals would use this new-fangled technology called SMS, or maybe the really fancy one called email, to let patients know when they are running behind. I just had to wait an hour and a half to be seen by the ultrasound department at my local hospital because of first scheduling conflicts, then just forgetting about me because they are so incompetent at management. If I came in an hour late, let alone an hour and a half, odds are I wouldn't even be seen. If I can't be seen on time, I d
Re: (Score:2)
Re: (Score:2)
We have that too. It all comes down to money and time to implement.
Re: (Score:2)
Most cases of sepsis occur from infections that the patient presents with.
Re: (Score:2)
Likewise [ibm.com], though the project is still in the early phases.
Re: (Score:1)
Re: (Score:2)
You understand the challenges with accuracy for such tools in the ED setting then(both practical and political)...I can only speak for my experiences so YMMV. I have seen predictive alerts for the ED be based on whatever seems to work well for the IP part of the hospital. IMO this can lead to poor sensitivity since you are working on fewer data points than IP might expect to see....Then there is the whole fact that no one wants to potentially miss a sepsis case so the home grown tools will probably err on t
Dr. Algo R. Ithms (Score:1)
With algorithm in the title (Score:2)
I really thought there was going to be medical blockchains somewhere in the story. Maybe next week...
Wait...wasn't Watson doing this? (Score:2)
https://www.youtube.com/watch?... [youtube.com]
And that video was from 2013.
But really, aren't "algorithms" what are used by humans anyway? Input data, apply logic and other constraints, eliminate some options, rinse and repeat until a 'best course of action' is shown? Or, is Watson "AI" and thus totally different? Or, is this a matter of deep learning, with blockchain technology being integrated next quarter?
I know buzzword bingo is nothing new, but it really, really feels like nobody remembers anything anymore.
Re: (Score:2)
I know buzzword bingo is nothing new, but it really, really feels like nobody remembers anything anymore.
It's social erosion. Random technical words or concepts are appropriated by the mainstream media and systematically bastardized: coding, AI, algorithm, hacker, cryptocurrency, etc. It's like when grannies and fat chicks start wearing the same thing as hot chicks, it's a death sentence for that specific trend.
Nope. Its a waste of money. (Score:1)
"Traditional methods such as monitoring hygiene and warning signs often fail to stop the disease."
THAT is why this won't work. This "solution" does not address THAT. This "solution" does not fix THAT. So no, it isn't gonna do a damned thing outside clinical trials.
On the other hand, when hospitals stop abusing antibiotics, and start actually maintaining cleanliness among hospital staff, infections go down, supebug development goes down, and hospital stays shorten, allowing the hospital to serve more patient
Re: (Score:2)
I've always wondered about that. Are hospitals (as in individual staff) competent and diligent in basic hygiene? I know they show all kinds of scrubbing in TV shows, but I see people in scrubs all the time on public transit and out and about. Are they all coming off shift?
Re: (Score:2)
Are hospitals (as in individual staff) competent and diligent in basic hygiene?
Some are, some aren't. I recently saw a video about someone who got a MRSA infection because a nurse didn't wash her hands. Apparently doctors tend to be pretty bad about washing hands in hospitals, though there's been a push for better hand washing or at a minimum disinfecting.
Here's my algorithm (Score:2)
Stop putting lots of sick people in a same building. Hospitals are becoming like those high-density pig or chicken farm where the animals are injected with antibiotics because it's cheaper than cleaning their shit. Smaller clusters = less problems.
Re: (Score:2)
Stop putting lots of sick people in a same building.
You want to have lots of little hospitals instead? Or have disastrously sick people just stay home? Or what?
How to deal with 'infections" (Score:2)
2. Clean wards and equipment. Have systems in place to ensure that is always done.
3. Dont allow your nation to deal with really high risk patients. Have a centre for tropical medicine ready to accept for the really interesting people.
4. Stop bringing really sick people into your nation. Have a visa system that demands medical results before a person gets to enter the nation.
List the
I worked in a hospital (Score:3)
Details:
KW Hospital - Kitchener, Ontario, Canada, laundry department
Years - 1987 -1989
Faulty practices - putting clean laundry on dirty laundry carts, staff covered in filth handling clean laundry, staff covered in filth delivering laundry, no fire or safety training (7 high school students got left in the building during a fire), no metal detector for sharp objects.
Algorithms like (Score:2)
quicksort?
(That got really dark)
Re: (Score:1)
Negative. Heapsort is the preferred sorting algorithm for hospitals and grave-diggers. They make a heap of dead bodies in O(n), and then they can extract the most decayed body in O(log n), repeating until there are no dead bodies left in the heap.
Re: (Score:2)
Far cheaper to just stop serving sugar additive (Score:2)
https://science.slashdot.org/s... [slashdot.org]
Recent development has indicated a popular "harmless" sugar additive as a likely culprit of causing two explosions in the occurrence of two nasty infections. Clostridium being one of them.
Start tackling that shit as prevention.
And yes, hand hygiene helps a lot, but is hard to do, as you would need to wash (with soap, not just alcohol) 100 times a day. That would cause a severe disturbance in the biotope on the nurses/doctors hands by itself!.
Re: (Score:3)
https://science.slashdot.org/s... [slashdot.org]
Recent development has indicated a popular "harmless" sugar additive as a likely culprit of causing two explosions in the occurrence of two nasty infections. Clostridium being one of them.
Start tackling that shit as prevention.
And yes, hand hygiene helps a lot, but is hard to do, as you would need to wash (with soap, not just alcohol) 100 times a day. That would cause a severe disturbance in the biotope on the nurses/doctors hands by itself!.
Yep, and not just sweeteners. 20% of C-diff conditions are caused by antibiotics. Anyone prescribed antibiotics should also be taking precautions to prevent C-diff taking hold.
Additionally, most people in he west don't get anywhere near enough fibre in their diet. Simply taking pre-biotic supplements (specific types of fibre) provides an environment in the intestines that encourage healthy bacteria to thrive and makes it extremely difficult for infections like C-diff to take hold.
Simply changing how doctors
Re: (Score:2)
Re: (Score:2)
Good job I don't think that rainbows and unicorns have any measurable effect on C-diff infection rates then ;)
Re: only upfront costs, the trouble with trawling the available data to look for significant patterns is that the available data may not be relevant or an effective proxy (metric) for what you're looking for: the so-called streetlight effect https://en.wikipedia.org/wiki/... [wikipedia.org] Identifying and collecting data that is valid and reliable is a skilled, expensive, and on-going process, not a one off expen
Re: (Score:2)
Re: (Score:2)
I know how ML works and you don't have to train the system forever.
That may work on complicated systems but not on complex adaptive systems. One of the characteristics of a CAS is extreme sensitivity to initial conditions. Any interventions to change patients' habits and behaviour will more than likely have profound and unforeseeable effects on the system (AKA "the butterfly effect") and which metrics are valid for detecting problems will change and so you end constantly looking for which metrics are valid proxies, especially if the interventions have been successful in en
Today it's "algorithms" ... (Score:2)
... next up: "blockchain."
Guilty feet ... (Score:2)
... and Al Gore have got no rhythm.