“This flu season is the worst in nearly a decade.”
The headline in the New York Times last week confirmed what even the most casual observer already suspected: that North America, and particularly the continental United States, is being gripped by an immensely challenging flu season.
The Centers for Disease Control and Prevention (CDC) estimates that hospitalization rates for the flu this year will top those from the high-severity 2014-2015 season. Over 710,000 Americans were hospitalized during the 2014-2015 season and 56,000 deaths were reported.
On January 26, in a recorded media statement, the CDC reported seven pediatric deaths, bringing the season total to 37.
“We’ve experienced two notable characteristics of flu this season,” said Dr. Dan Jernigan, Director of the Influenza Division in the CDC’s National Center for Immunization and Respiratory Diseases during the media session. “The first is that the flu activity became widespread within almost all states and jurisdictions at the same time. The second is that flu activity has now stayed at the same level for three weeks in a row.”
Currently, healthcare professionals and organizations are pouring their energy into the most important task at hand – managing outbreaks, putting mitigation strategies in place and working to keep staff and patients safe.
However, as the urgency of the flu season reaches a pitch, experts are already working towards a future where modern technology could change how healthcare organizations prepare for the flu.
The prospect of forecasting and machine learning
What would you do if you knew what was coming?
Experts in machine learning are shifting the flu conversation to the idea of forecasting, and the promise it holds for healthcare organizations in the future who are looking to get a head start on the flu before patients start flooding in with sniffles, fevers and sore throats.
Flu forecasting isn’t exactly a new concept. Each year predictions are released, in fact for the last five years the CDC has facilitated a flu forecasting challenge, recruiting the best minds from universities, corporations and labs to take their best shot at predicting what the season holds.
However, the thirty-plus organizations involved in the challenge rarely reach anything close to a consensus. Some experts attribute this to the scale on which the predictions are made. Current forecasting and tracking is done by aggregating high level data, often by region or geographical area. This year was the first that the CDC has drilled-down to the state level and provided it to those in the forecasting challenge.
Each year, healthcare organizations collect a wealth of data on the flu – number of cases, number of hospitalizations and deaths, transmission patterns. Combined, this data holds a wealth of predictive insights. Machine learning is one way to shift through that data and learn from historical patterns.
Though looking to the past to predict the future, Rosenfeld advises that the next step is to examine the data on a much smaller level. The flu hits cities, towns and even individual hospitals very differently, reducing the likelihood that predictions made at the region or state level will be applicable to all healthcare organizations and communities within it.
During the media conference, Dr. Jernigan estimated that we’re about halfway through this flu season, meaning that despite signs that flu activity has peaked in some areas of the US, there is still more yet to come.
The question that remains for experts focused on the future of the flu is how technology can shape the way the flu is forecasted, and how a more contextual level of surveillance could one day turn the tables on seasonal illness.
Learn more about what’s in the works with RL6:Infection’s context based syndromic surveillance in the whitepaper A New Perspective on Infection.
About the Author
Samantha is part of the marketing team at RL and is passionate about sharing healthcare stories. When she's not typing away, you can find her as far from the city as possible with a book and a kayak.More Content by Samantha Relich