eVision: Influenza Forecasting Using CDC, WHO, and Google Trends Data

Abstract

Influenza, more commonly known as the flu, is a contagious respiratory illness caused by viruses which in the 2018–19 flu season, infected 37.4 to 42.9 million people in the United States alone. Of those, 431 to 647 thousand were hospitalized and 36,400 to 61,200 (most of them elderly and children) succumbed to the disease. At the time of this writing, the best known defense against influenza is vaccination. However, due to the annual mutation of the very many strands of the flu virus, new vaccines must be administered every flu season. Hence, the prediction of the rate of growth in reported infection cases of each strand of the flu is paramount to ensuring the correct supply of vaccines per strand. Machine learning - specifically Neural Networks - are a great tool for making future predictions using existing data. Long Short-Term Memory (LSTM) neural networks are utilized by Santa Clara University's EPIC (Ethical, Pragmatic, and Intelligent Computing) and BioInnovation & Design laboratories for continued research and development of a tool named eVision (Epidemic Vision) to predict the trend of influenza cases throughout the flu season. eVision has been trained on data gathered across 4 flu seasons from the 2014–15 season to the 2017–18 season of the Center for Disease Control and Prevention (CDC) records as well as the World Health Organization (WHO) and Google Trends search result data gathered across the same period of time. eVision has been able to make 7 weeks in advance predictions about the flu trend in the 2018–19 United States flu season with 90.15% accuracy. This paper is to report the achievements of eVision thus far and to delineate next phases for the project which aims to provide a tool for the pharmaceutical and healthcare industries to more accurately predict the trend of flu (and other) epidemics in order to meet the demands for vaccines and test kits ahead of time.