Each year, the flu is responsible for the deaths of between 3,000 and 49,000 people in the U.S. alone. An accurate flu outbreak forecast could have a significant impact on the way society prepares for the flu.
The Center for Disease Control, along with the rest of the world, is looking for a better way to forecast flu trends in order to predict and prevent significant infection rates. Currently, flu outbreaks are monitored by a method that relies on public health officials to voluntarily report the percentage of patients they see with flu-like symptoms like a temperature higher than 100 degrees, a cough and no other explanation. Enter the Internet.
The current method of monitoring the epidemic is antiquated and greatly subject to human error. While these numbers might give us a sense of when a flu epidemic is eminent, it doesn’t account for those doctors who simply don’t have the time to report or for the fact that they may be reporting as they get around to it – after seasonal peaks. During an actual flu outbreak they may simply not be able to take the time and talk about the flu to the CDC because they are busy treating actual patients.
These voluntarily reported statistics couldn’t be 100% accurate or even close because they don’t take into account people who have the flu but don’t go to the doctor for treatment.
Additionally, the network that reports this data is slow. It averages two weeks for the numbers to filter through the system so the data is never current. That’s two weeks after the data is reported to the network, which could be two weeks after the data was actually collected which means, in theory, the data could easily be a month old.
One new trend in monitoring the flu is by monitoring Google and other search engines to see when searches for flu and flu-like symptoms are being highly searched. In this day and age, it stands to reason that many people head straight for the internet to research symptoms before ever considering calling their doctor. It’s a good idea in theory; if there is a high search volume for these symptoms then there is probably an outbreak.
The approach has had some success but also one monumental failure. Last year when the flu epidemic hit, the media wrote about it and then people began to search for it. This is when the failure happened. From the amount of people searching for the flu and flu-like symptoms, Google Flu predicted a huge flu outbreak that never came to fruition. Meanwhile, the media wrote about the impending outbreak using the Google Flu prediction engine as its source. Basically, Google skewed it’s own data by being so highly searched.
Google flu also only looked at the previous year’s CDC data once and never checked back nor did it check its predictions for accuracy against the CDC’s reports for the current year. Basically, it was a great idea that ended up being a guess because the predictions were never confirmed against actual flu cases.
Waiting for doctors to have the time and gumption to report flu outbreaks is not a realistic answer, and while Google is on the right track, there needs to be a system in place that consistently flags flu and flu-like symptoms in a patient’s electronic chart and automatically reports them directly to the CDC, removing the burden of documenting data from health care providers.
Predicting flu outbreaks in real time based on search engine results may be more accurate than waiting for health officials to report symptoms but both come with a high probability of human error. We’re on the right track but we need something more predictable and less prone to be misreported.