A technique called "nowcasting," which combines two forecasting methods with machine learning, can be used to track flu activity locally.
Researchers from the Computational Health Informatics Program at Boston Children's Hospital developed the technique, called ARGONet. They published details of the technique and its efficacy in Nature Communications.
The ARGONet technique combines two flu detection models:
• AutoRegression with General Online information model, which leverages information from EHRs, Google searches and historical flu activity in a certain area
• A model which draws on spatial-temporal flu patterns in surrounding areas
The models were combined with a machine learning system that processed flu predictions from both models as well as flu data.
Researchers applied the ARGONet technique to flu seasons from September 2014 to May 2017.They found the technique was able to make more accurate predictions in 75 percent-plus of the states studied than using the ARGO forecasting method alone.
"Timely and reliable methodologies for tracking influenza activity across locations can help public health officials mitigate epidemic outbreaks and may improve communication with the public to raise awareness of potential risks," said Mauricio Santillana, PhD, a faculty member of the Computational Health Informatics Program and the study's senior author.