Slashdot Mirror
Google Algorithm to Search Out Hospital Superbugs
Googling Yourself writes "Researchers in the UK plan to use Google's PageRank algorithm to find how super-bugs like MRSA spread in a hospital setting. Previous studies have discovered how particular objects, like doctors' neckties, can harbor infection, but little is known about the network routes by which bugs spread. Mathematician Simon Shepherd plans to build a matrix describing all interactions between people and objects in a hospital ward, based on observing normal daily activity."
Markov chains are the original, mathematical theory. This is just an application of Markov chains to tracking disease transmission, a fairly common method that long pre-dates Google. Google's page rank algorithm is another application of Markov chains to citation ranking (and, as it turns out, it wasn't the first time that it was applied to that either).
In 2005, Britain's going nucking futs over MRSA. It was used as a reason to justify taking the NHS (National Health Service. Translation: universal healthcare) and molding it into whatever each Party wanted the world to be like. You couldn't pick up a newspaper without SuperBug this or SuperBug that on the front page.
Meanwhile, in America, the sound of crickets gently chirp. Chreeeep, chreeeep, chreeeep. Nobody gave a tinker's cuss about MRSA. At all.
OK. That's the scene. People in Britain thinking that MRSA is going to turn the country into 28 Days Later. America thinks MRSA is some rapper's name.
And then the official numbers came out for MRSA deaths for that year.
England/Wales, in 2005: 1629 deaths.
United States, in 2005: 18,650 deaths.
There are more people in the States than England and Wales. So I looked up the numbers for the land of the free and the home of the Whopper and Pommie/Limey/Rosbif-TaffyLandSheepCountry.
US population at the time - 295 million.
Eng-Cym population in the last census (and it won't have doubled from 2001-2005) - 52 million.
So what were the chances this would have killed YOU? Well, remote (if you're reading this now), but what about back then? The equation is:
[population of the country in 2005] / [deaths from MRSA there in 2005]
= [chance of being killed by MRSA in 2005].
The chances you had of MRSA killing you in England and Wales, with everyone going mental over it, in 2005 - 1 in 32,000.
Chances of dying the same death in a country with market-driven health system, where people are NOT specifically looking for MRSA - 1 in 15,800.
I'll let those numbers sink in. British readers might want to look at them again and make sure up is still up.
And now I'm going to pretend to be really stupid here: I could be spectacularly wrong, but it LOOKS like the numbers prove a person was twice more likely to kick the bucket from MRSA in the States than in Blighty (OK, England and Wales. I'll let someone else add Scotland and Northern Ireland to the mix). America, with its pay-as-you-go health system making monster profits, not as good as a system some people would tell you is on its last legs.
What was even funnier (maybe 'funnier' isn't quite the right word) was the excuse used in the UK National Statistics Office for why their number was so HIGH:
This is either the longest and most researched Flaimbait ever to appear on SlashDot, or I just blew. Your. Freaking. Mind.
Unless you're American: in which case, just think of this like the slang you don't understand in Doctor Who, words like 'chav' and 'ASBO'.
Shiny. Let's be bad guys...
If it's really like Google, then no doubt the Wikipedia page on Tuberculosis will also be gamed up higher in the rankings than any real scientific paper or expert description of the bacteria.
The Wikipedia page will probably be no more than a few paragraphs. There will be at least 3 flagging boxes (declaring it to be a stub, not citing sources, containing trivia and that trivia is Verboten and punishable by death!) There will be at least one link to an article about a band you've never heard of who once wrote a song about tuberculosis. The talk page will be 10 times longer than the article itself and have at least one sentence praising Ayn Rand.
Sounds like you just described the PageRank TM algorithm in general. Remember, the "Page" in PageRank TM is for the founders name, not web pages. It is not the complexity of the math, but of the ability to solve equations on a large scale quickly over the distributed systems that makes the Google methods so powerful in solving these "simple" math problems.
If only it were as simple as you say!
Staphylococcus aureus lives on normal skin. You're probably crawling with billions of these little buggers. Such is life. Most of the time, methicillin-sensitive Staph outgrows the methicillin-resistant Staph (in theory, there is a difference in rate of replication, since MRSA has that extra cassette that it needs to copy)
But in certain places in the U.S., community-acquired MRSA infections make up as many as 35% of all infections (from simple skin infections to bacterial sepsis), at which point isolation practices are pretty pointless, particularly since they've never been proven to actually decrease rate of transmission. (Although granted, if you die from MRSA, it's going to be hospital-acquired MRSA that gets you.)
Other multi-drug resistant bacteria are prevalent in the environment--in the soil, on flat surfaces, in computer keyboards--and don't cause illness in immunocompetent people. Examples are Pseudomonas aeruginosa, Acinetobacter baumanii, Enterobacter cloacae, Stenotrophomonas maltophila, and Alcaligenes xylosoxidans. But Gram-negative rods, more so than Gram-positive cocci, are likely to kill you in less than 24 hours. (Necrotizing fasciitis--caused by so-called "flesh-eating bacteria"--is more likely caused by Group A Streptococcus, which is very sensitive to straight-up penicillin, and MSSA)
MRSA and other multi-drug resistant bacteria are simply not transmitted by air. Respiratory isolation rooms (negative pressure rooms with separate air filters) are good for preventing the spread of pulmonary tuberculosis, and various non-fatal viral illnesses, but that's about it. And when you've got 30-40 kids with RSV, there's no way you're going to isolate all of them.
Most vectors of these bacteria are not sick from them, and they're so prevalent that it's not practical to exclude people who are colonized with this stuff. Most health-care workers are definitely colonized, for example, and it's no good preventing these people from working. Making people who have upper respiratory symptoms wear masks may be helpful, but making everyone do so? Studies don't show any difference in transmission.
Any linens that might have been contaminated are destroyed anyway.
The measures that have been proven to decrease transmission of bacteria and viruses are (1) thorough hand washing, meaning lathering up and running your hands under water for at least 15 seconds or (2) using the alcohol-based anti-bacterial gels (although this won't destroy Clostridium difficile spores) Everything else (masks, gowns, gloves), in terms of preventing transmission of these bacteria, is, according to the studies, infection control theater. (I'm not talking about universal precautions here, which definitely keep health care workers from getting HIV and hepatitis.)
And when you come in on a backboard with C-collar because you were in a car-crash, infection control sort of takes a back seat until they take care of your airway, breathing, and circulation. There's no way you're going to keep someone dying from trauma in a quarantined area until you make sure they're not colonized with any of this stuff.
Bottom line: wash you hands, and stop asking your physician for antibiotics whenever you or your kid have the sniffles. That'll cut the incidence of MRSA and other multi-drug resistant bacteria.