Web-Crawling Program Spots Disease Outbreaks

no1home writes "There is a story at Discovery Channel's site about a new utility for mapping disease. The premise is to have bots crawl the web looking for stories about disease outbreaks and log them onto a map. '"We were originally thinking about how we could expand disease surveillance and pick up outbreaks earlier than traditional methods," said John Brownstein of Harvard Medical School and Children's Hospital Boston, who created HealthMap in September of 2006 with Clark Friefeld, a software developer at Harvard Medical School.' But then it was noticed by Google.org and has since grown into its own website, HealthMap Global disease alert map, and claims to be able to identify 95% of all disease outbreaks, some of them before WHO or CDC."

Great!

Now all the hypochondriac hyper-nerds have another reason to sit home on their computers, cloistered from the outside world. :-)

Low on the Useful Meter

Well, it's not really a "map of disease"
breakouts. In fact the map part is rather just
a shiny pony?

A list could have done just the same amount
of good. Since for the most part each area has
one pushpin that just sums up the area.
[FWIW, I only looked at US pins.]

I was expecting a cluster map, like you see on...
Wunderground Wundermaps [wunderground.com]

or on...
http://www.housingmaps.com/ [housingmaps.com]

At least if it was a cluster map I could
look at an area and think, "I sure as heck
ain't traveling there for

Somewhere in Madagascar..

CLOSE THE PORTS [quartertothree.com]

Re:

How about a bot that spots new memes by checking Wikipedia articles for repeated vandalism and sudden article protection.

Catch the video

Fascinating TED Talk [ted.com] on a similar (the same?) project? As I recall, some of video was a bit unpleasant to watch, but (IMHO) very worthwhile.

I still think

a facebook or other social app for people to self report symptoms is a great idea that no one has uilt yet. one could even "out" symptoms of their friends or speculate which friends made them sick. lots of issues with it, but a different data source for inf disease folks, even if the data was not completely accurate, would be helpful in predictions.

too busy to do it myself now...

Just what the insurance companies need

> one could even "out" symptoms of their friends or speculate which friends made them sick. lots of issues with it, but a different data source for inf disease folks, even if the data was not completely accurate, would be helpful in predictions.

Yeah, right, just what we need, an inaccurate resource for the insurance companies to data-mine. Your premium has now increased by a factor of 5, just because someone with your name (Mike Smith) allegedly made someone else sick. Great.

No thanks.

Re:

Today
Jenny Smith gave you the clap! Give her measles? 3:56pm

Re:

Better still, look at what symptoms people are searching for. Sure, you'll get a little noise every time a repeat of "House" is aired. However, when google gets systemic swaths of "butt bleeding" , "grey vomit" and "ocular hemorrhoids", bad things might be coming.

This thing has one flaw

as far as I can make out. It relies heavily on human reporting. And sometimes it takes a while for news on disease outbreaks to make the news.

Unless there is some way to report directly TO this crawler, I seriously doubt the claim that a web crawler can know of outbreaks before the WHO does.

hmm... I just referenced The Who - a band...

Re:

You are indeed correct.

I forgot one thing: there is more to news than the mainstream media.

A local blog could contain information about an outbreak of disease days before the WHO (again with the band reference) or CNN/BBC finds it advertising revenue generating worthy.

Neat

This is neat - although how useful it is I don't know... It'll be pretty obvious that communitites not tied to the www 24/7 will be sorely under-represented. Also, the disease categories seem a bit narrow - it would be cool to have stuff like murder, violence and alcohol related-disease in there. Ok, not transmissible diseases in their own right but they still have some pretty profound health-related consequences.

Re:Neat

"It'll be pretty obvious that communitites not tied to the www 24/7 will be sorely under-represented."

And those are the communities which have the highest outbreaks of disease... So it seems pretty pointless to me.

Usefulness?

The CDC, and local and state health departments all have a list of "reportable" diseases. (Things from TB to gonnorhea to ebola to SARS) If a doctor encounters them, they are supposed to notify the health authorities. That is for biostatistics and epidemiology purposes.

If they have to look these cases up in the news instead of getting notified by hospitals and clinics, then the system is in a really bad shape.

Re:

If they have to look these cases up in the news instead of getting notified by hospitals and clinics, then the system is in a really bad shape.

Very true - but might there be value in understanding the public's awareness of disease? One thing that this map might measure is a communitites awareness of transmissible disease and awareness *should* lead to protective behaviour. So if there's a mismatch between regular epidemiological stats and this map then perhaps public health bods should going in there telling people to wear condoms, wash their hands, etc.

Re:

The CDC, and local and state health departments all have a list of "reportable" diseases. (Things from TB to gonnorhea to ebola to SARS) If a doctor encounters them, they are supposed to notify the health authorities. That is for biostatistics and epidemiology purposes.

What about non-reportable diseases? German Measles, Chicken Pox, and many others are not reportable, and most people wouldn't even bother going to the doctor if their kids came down with them (or is that not the case anymore? seems like every

What do they expect from this..

What the designers expect from this will highly color my opinion on it. The article linked isn't exactly clear on this.

Do you want to track and try and predict disease breakouts in first world areas then probably decent, track world wide stuff then terrible. Outside of the obvious (self reporting) there is the whole issue of how much of the world is on the internet? While much of the first and even quite a bit of the second world countries are on the vast majority of the population doesn't have computers, let alone internet access.

I can easily see many many great uses for this and I expect all of them to be explored at some point - I can also clearly see many not so great uses and I fully expect them to be used too. As the old saying goes, there are lies, damn lies, and statistics.

In fact we can already see the maps being posted and used by people who have little to no understanding (if we are generous, I'm sure some understand and use them to further their own aims) to say things the data *can not say* and it isn't even mainstream yet. *sigh* It's like many things we have today - the greater amount of good it can do the greater amount of abuse one can use it for too.

I never would have guessed

Apparently news reporting is still good for something. I never would have guessed.

Missing important diseases...

I was looking forward to viewing the "erectile dysfunction" map based on viagra posts.

wasted effort

pathology labs already have a list of reportable diseases that the CDC monitor (you know, their job).

why would anyone rely on reports from the media on what outbreaks are going around when you have trained professionals with lab equipment diagnosing these illnesses to begin with?

Eek.

I've been playing Pandemic 2 [crazymonkeygames.com] all night and this is really freaking me out.

I think I'm moving to Madagascar...

Another method...

The New York City Dept of Health monitors sales records of certain medications gathered from drugstore chains to detect disease outbreaks and biological attacks.

Dupe

Another report of the same HealthMap thing was on /. not too long ago:

http://science.slashdot.org/article.pl?sid=08/07/09/1424247 [slashdot.org]

and what I said then still stands - the plural of "anecdote" is not "data". I defy anyone to come up with useful statistical models and tests on actual disease incidence based on web-crawling for disease names.

Hasn't this story been covered already?

I believe we just saw this. A search on this site for the single word "disease" shows this link 3rd:

http://science.slashdot.org/article.pl?sid=08/07/09/1424247 [slashdot.org]

No google foo required to find this dupe.

jeff

Calling bullshit... hello... bullshit do you hear?

It is not that I don't like the idea but it is essentially flawed.

A) It still requires human input. No one reporting the disease does not mean that it is not there.
Looking at former Yugoslavia and seeing only 1 case of meningitis while here in Bosnia everyone knows about (and it is on TV, radio and in the papers) the brucellosis [wikipedia.org] epidemic that has been going on for months or even years maybe.

B) That input must be made over the internet.
Look at Africa. It is practically squeaky clean. There is one case diarrhea in the entire Botswana. And everyone is completely healthy up in the North.
Could it possibly be due to the lack of internet-based inputs instead of due to the lack of diseases?
Check out UK or the East Coast of USA. They are crawling with diseases.

C) It should preferably be in English. Can the crawler read any of these articles:

http://www.zzjzfbih.ba/content/view/66/13/ [zzjzfbih.ba]
http://www.dw-world.de/dw/article/0,,3259389,00.html?maca=bos-rss-bos-all-1475-rdf [dw-world.de]
http://www.slobodnadalmacija.hr/BiH/tabid/68/articleType/ArticleView/articleId/14733/Default.aspx [slobodnadalmacija.hr]
http://www.dnevniavaz.ba/dogadjaji/panorame/bruceloza-prepolovila-prodaju-livanjskog-sira- [dnevniavaz.ba]
http://www.blic.co.yu/repsrpska.php?id=44508 [blic.co.yu]

Basically, what they come to is that there is a SHITLOAD of cases of brucellosis among the various cattle in Bosnia.
And that it is going to stay that way for a long time, cause nobody is really doing anything about it.

It is a fine idea, but unless you have every square kilometer of Earth covered with internet access and people who will report it in a language that the crawler understands - it is beyond useless.
Even dangerous.
Zoom out over Asia and turn on the Google in Chinese under Feeds. China's disease count jumps from around 40 to around 140.

No. You can't fix all the problems by "putting it on the internet".