Slashdot Mirror
When Google Got Flu Wrong
ananyo writes "When influenza hit early and hard in the United States this year, it quietly claimed an unacknowledged victim: one of the cutting-edge techniques being used to monitor the outbreak. A comparison with traditional surveillance data showed that Google Flu Trends, which estimates prevalence from flu-related Internet searches, had drastically overestimated peak flu levels. The glitch is no more than a temporary setback for a promising strategy, experts say, and Google is sure to refine its algorithms. But with flu-tracking techniques based on mining of web data and on social media taking off, Nature looks at how these potentially cheaper, faster methods measure up against traditional epidemiological surveillance networks." Crowdsourcing is often useful, but it seems to have limits.
Modern epidemics and pandemics are almost ALWAYS overestimated by those predicting them. In part, this is because those predicting them often have a vested interest in making them sound a scarier than they actually are. So you get a lot of this "The sky is falling! Weessa all gonna die! Give me more research money!" screaming from epidemiologists and those in related fields.
What political party do you join when you don't like Bible-thumpers *or* hippies?
The same idea can be used to find mass shooters before they fire a bullet. We'll start arresting people based on their search and CC usage history. And mainstream America will be happy "because we're safer".
the joke was "I opened the window and influenza"
Computer modeling is a powerful technology that should not be underestimated.
However, it should also not be overestimated.
When the "real world" has millions of convergent factors responsible for an event, computer models can sometimes capture a few thousand. Based on those, a simulation is created that suggests a certain outcome. But it may be using less than 1% of the necessary data.
This is like making architectural models out of child's blocks and then being surprised when the building falls down after it is eventually made. There are issues of scale in addition to data that can reveal periodistic or epicyclic patterns that cannot be modeled in a linear method.
Financial incentive? In science?
Well, yes. Scientists are people too, and they want the same thing most of us want: to put together enough of a money pile to leave the rat race adn go do what we want for a change, without having to make it profitable and thus bending it to the lowest common denominator (LCD).
Michael Crichton's State of Fear reveals this tendency in our media and science. Quite simply, fear sells. And what doesn't sell will not get funded, will not help your 401(k) swell, and will leave you an unethical but underpaid lab-drone while fools get the gold for preaching what people want to hear.
Distorting the whole thing even more.
It's useful, but obviously not 100% reliable. How that's not clear to everyone is not clear to me.
drastically overestimated peak flu levels
So do most "professionals" that study it.
H1N1 was so incredibly over-hyped, for example.
What do I know, I'm just an idiot, right?
They should subtract out a factor based on how much the flu is being talking about in the media.
In short, a system that learns from abnormal circumstances will no longer work as well under normal circumstances. This year's flu outbreak didn't follow previous models, so Google's application of those models was inaccurate... but we'll blame Google for it anyway, and cast shame upon them for being so terribly wrong.
Of course, the article is much better, delving into other systems that also predict and monitor flu outbreaks, and why they were or were not correct. TFA is really about the difference between traditional reporting sources (as from doctors' offices) and newer data-mining approaches (harvesting from searches and Twitter).
Screw you, Slashdot.
You do not have a moral or legal right to do absolutely anything you want.
Google won't ever get this right if they can't figure out where a user is. For example - we have 6,000 in a campus environment in a city in the San Francisco, CA Bay Area. We are behind a proxy server located right on campus. However Google thinks we are in some Latin American country (you can tell when you can't order via Google play and when devices show up in the USA and you can order them from home but can't order them from work because "it isn't available in your country" shows up). So if we search for Flu, Google things someone in Latin America has the flu. They would get better results if the geo location code worked better.
Why stop there? Just arrest people for non-conforming behavior.
Anything but shopping, going to work, watching TV and loudly proclaiming "they hate us because of our freedom, liberty, peace, diversity, consumerism, sexual liberation and excellent shopping" is suspect.
If we round up these deviants, I think we can achieve Utopia within the decade.
This is probably because people will update their social media sites with claims of having the flu. If they actually had the flu odds are they would not have the strength to even do that.
The real flu is pretty terrible and people often think they have it when they have a minor cold.
"Good morning, gentlemen. What does the overnight Google search analysis show?"
"Well, there is the continued flu outbreak on the east coast, with the biggest concentration in Boston. There seems to be a ringworm outbreak in pets in the southwest, and our numbers show, and I caution you this is probably a 60% overestimate, the apparent nationwide removal of 3.8 million brains due to unspecified causes."
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
It's only a matter of time before a real flu epidemic rages though the world. The trick with flu is the balance between it's virulence and it's morbidity. Flu's that come by that are virulent AND overly morbid will burn out. People will die too fast to spread the disease. This is why there has been no world wide outbreak of Ebola... it kills so fast, it can't spread. A mild flu (low morbidity) can spread far and wide, because it doesn't kill the majority of its hosts, thus allowing them to pass the disease on. But eventually, you'll get another 1918 flu, that is easily transferred AND has high morbidity. When that happens, we'll be better prepared in that we have drugs like Tamiflu now, and also have antibiotics that stop secondary infections like pneumonia. Those will moderate the disease in the first world, but the third world will still have results like 1918. The flu will still only have a year to do its nasty business, as a vaccine will undoubtedly be developed and administered. But that takes a year to do and there is no good way to speed that up. So at some point, we'll get a scenario like the one portrayed in Contagion, which was an excellent film in my estimation, showing what a 1918-like flu epidemic would look like today. If you haven't seen it, you should. It shows how the government won't be there to help us in the short term at the very least, and why it is important to be able to be relatively safe when isolating yourself for a long period of time (up to a year)... Prepping isn't crazy... it's common sense.
Brawndo: It's what plants crave!
Google announces they're tracking the flu (hey everyone, come see a map that will tell you how bad the flu is in your area!), Larry Page announces he's offering free flu shots to all kids in the Bay Area, and Google announces it's launching a flu shot locator. Of course searches for "flu" and "influenza" are going to increase. That will throw off the accuracy of your model. What they're really measuring is this: "people who are thinking about the flu and proactively reaching out to learn more."
...when people regularly died from it?
0 1 - just my two bits
Of course the sensational news story of this past winter was the rampant outbreak of "flu" which suddenly has become one of the biggest health scares the world has ever seen.
Google needs a sensational hyperbole filter on their Internet scrapes, something to blow past the kind of rampant proliferation of "news" not based on fact or reality, but only reported to drive web hits or broadcasts has become common place these days. Some reporter goes to the ER of a hospital, sees a room pack of sniffling, coughing people in the middle of winter, and then declares there is a plague of epic proportions infecting America, which is then dutifully re-tweeted to the idiot masses.
First, its just Flu. While very young or very old might be prone to complications from flu, the vast majority of people getting flu WILL NOT DIE from it, so the fear and overreaction to flu is unwarranted.
Second, nothing has changed from, say, 20 years ago. There is no "rise" in flu infection rates, there is no epidemic. Flu isn't stronger or more powerful then it used to be. The only difference is the Internet which spread misinformation causing ignorant alarmists to jump to ridiculous conclusions. Was 2012-2013 record breaking for flu outbreak, or was it just the year that the social media reaction to an otherwise normal amount of flu infections reached epic proportions online?
People get sick and if you eat healthy and take your vitamins generally you can recover from it easily and quickly. I have survived 2 rounds of flu spreading through out my place of work without getting sick and I haven't even gotten the flu shot, I just up my vitamins and drink some more OJ when everyone else around me is coughing and sneezing their snot all over the place. However, I'd rather people just stay the fuck in bed when they are sick.
I haven't thought of anything clever to put here, but then again most of you haven't either.
What were there, like 6 people who died from it last year in the US?
but apparently they have a whiz-bang hypochondria pandemic detector.
Welcome to the Panopticon. Used to be a prison, now it's your home.
Prediction is always in error, or you all would be a multibillionaire by now. So you can under predict, or over predict. As for prediction of spread of disease, which way will kill more people: under predict so that people don't take precaution, or over predict that people gets too careful ?
Personally I find the mania detector quite useful, and hope it can be used to expose other mass-illusions such as hybrid cars being positive for the environment and guns in school being a good thing. I know I won't need a flu shot, but I want to know how many crazy people I should prepare to disbelieve and avoid any given day.
I don't get flu shots either. Maybe there is a connection.
Be seeing you...
The project to promote panic over a potential deadly flu outbreak was well organised and well financed. Google was proud to be at the forefront of black-propaganda output. What would you expect from the most important asset the NSA has ever created? Google's hardware and software designs are found in 'shadow' Google systems used by all the major intelligence organisations of the West, to hold their massive databases, and to allow said information to be searched and mined.
The current 'flu' project began with the digging up of corpses from the 'plague pits' of the flu outbreak early in the 20th century that killed tens of millions. The corpses were retrieved to allow the genetic material from the historic flu viruses to be sequenced, replicated, and introduced into current strains of flu. It was these engineered viruses that were then released into the wild, with the expectation that they would cause a major deadly flu epidemic. Google merely predicted what its side expected to result from its work.
It is a fact of Human history that bio-weapons have proven to be singularly unsuccessful. The natural circumstances that allow a particular biological attack to be deadly to a population are not replicated so easily. The idea that the genetic code of a previous deadly flu would prove to be fatal today thankfully proved false.
Those of you stupid enough to think your masters mean you well will not comprehend why the American government in particular (but others as well) continue to research bio-weapons, and regularly release manufactured agents on unsuspecting civilian populations. Of course, these experiments usually involve 'harmless' or mostly 'harmless' agents, supposedly so their rate of infection can be tracked. However, for children in the care of the state, orphans, prisoners, and members of the armed-services, such biological experiments are frequently anything but benign. Only a few years back, New York proudly turned its entire 'orphan' population for involuntary AIDS research.
Google and Microsoft are owned and run by people who make regular statements about how the Human population of the planet must be massively reduced. Since the USA is the land of the eugenic policies and research that Adolf Hitler proudly adopted, no one should be surprised when Google continues this eugenic legacy. There would have been many in the UN and Google horrified when their engineered flu strains failed to have any major impact at all. They prepared the world for millions of deaths, but forgot about the concepts of 'evolution'. It is in the nature of life to out evolve particular strains of deady mutating infections.
Producing a guaranteed killer flu with new, untested genetic sequences is infinitely more difficult. They'll need to find another way.
hahahha yeah... how can you base a quantified number of people infected off of google searches... do you know how inflated the list of other conditions of interest could be if they applied the same techniques??????? dumb dumb dumb dumb dumb
Google Flu has never been used to officially declare a flu outbreak. It's a neat tool, and it has been successfully used in retrospective studies, but until it actually helps us prepare for a flu outbreak in ways above and beyond what traditional surveillance already does, it will continue to just be a neat tool and not a useful one. The same goes for the Twitter flu prediction models. These tools are cool, but unless people actually do things differently to prepare for an outbreak based on their predictions, they don't mean anything. Consider this question: If you were a public health professional and you knew about a flu outbreak 2 weeks earlier, what would you do differently? Encourage people to get vaccinated? Already being done. Shut down schools? You had better be damn sure. Warn local hospitals? You are kidding yourself if you think they are going to start bringing in extra staff in the hopes that your prediction was right. So really, what does that extra week or two get you?
Computer models suggest global warming a threat to human existence.
Google Flu is a joke and every Epidemiologist knows it. The CDC and every state and local health department gets their data directly from hospitals and doctor's offices.
The gap Google is trying to fill is of people that are not sick enough or refuse to go to a hospital. But all this can really detect is where the people are coming from.
In short: Obligatory