Smart Outbreak Detection Using Online Info
March 16th, 2009 | Published in Google.org
An interesting article making a case for the use of online information for earlier detection of disease outbreaks came out last week in the Canadian Medical Association Journal. The article, co-authored by HealthMap co-founder and Google.org grantee John Brownstein, looks at an outbreak of listeriosis in Canada and compares online search trends, news reports, and diagnosed case counts during the outbreak period.
Listeriosis is a bacterial infection often caused by contaminated food. The outbreak in Canada killed about 20 people last summer. By using Google Insights for Search data combined with HealthMap's news surveillance data, the authors discovered that a specific search query provided even earlier indication of the outbreak than news reports.
So, what was that key search term that gave the earliest indication of the outbreak? Listeriosis. That's right, when the researchers looked at trends for people searching online for the technical term "Listeriosis," they discovered that the peak of the search trend for that term was the same as the actual peak of medically diagnosed Listeriosis cases. The increase in online searches for "Listeriosis" began in mid-July, one month before the federal announcement that an outbreak was underway in Canada. This means that people diagnosed with Listeriosis, or others close to them, were likely to be the ones searching for that term online at the time of diagnosis, causing a spike.
In contrast, search trends for "Listeria," the term used in the public announcement about the outbreak, peaked around the time of the announcement and other news reports in mid-August. Thus, people searching for "Listeria" were probably doing so in response to the press about the outbreak, not because they'd been diagnosed.
Early detection is critical to helping health officials respond more quickly. While documenting the potential for using online info for earlier outbreak detection, the authors also recognize the challenges. This seems to be the tip of the iceberg for research in this area, and it has others talking too - check out the Wall Street Journal blog which calls this "a wonkier example of Google Flu Trends."
Listeriosis is a bacterial infection often caused by contaminated food. The outbreak in Canada killed about 20 people last summer. By using Google Insights for Search data combined with HealthMap's news surveillance data, the authors discovered that a specific search query provided even earlier indication of the outbreak than news reports.
So, what was that key search term that gave the earliest indication of the outbreak? Listeriosis. That's right, when the researchers looked at trends for people searching online for the technical term "Listeriosis," they discovered that the peak of the search trend for that term was the same as the actual peak of medically diagnosed Listeriosis cases. The increase in online searches for "Listeriosis" began in mid-July, one month before the federal announcement that an outbreak was underway in Canada. This means that people diagnosed with Listeriosis, or others close to them, were likely to be the ones searching for that term online at the time of diagnosis, causing a spike.
In contrast, search trends for "Listeria," the term used in the public announcement about the outbreak, peaked around the time of the announcement and other news reports in mid-August. Thus, people searching for "Listeria" were probably doing so in response to the press about the outbreak, not because they'd been diagnosed.
Early detection is critical to helping health officials respond more quickly. While documenting the potential for using online info for earlier outbreak detection, the authors also recognize the challenges. This seems to be the tip of the iceberg for research in this area, and it has others talking too - check out the Wall Street Journal blog which calls this "a wonkier example of Google Flu Trends."