Replication Data for: Using electronic health records and Internet search information for accurate influenza forecasting

The data for replication contain three parts: Centers for Disease Control and Prevention (CDC) data, Google Trends (GT) data, and Electronic Health Record (EHR) data. All data are obtained and frozen as of July 9, 2016. CDC publish weekly unweighted Influenza-like Illness (ILI) activity level (gis.cdc.gov/grasp/fluview/fluportaldashboard.html). The initial report is subject to revision in later weeks as more data are gathered and processed from participating clinics around the country. We have consolidated the CDC's weekly unweighted ILI activity level data with later revisions into one single csv file. Google Trends (www.google.com/trends) data are publicly available. The query terms that we used were identified from Google Correlate (www.google.com/trends/correlate), where we identified 129 flu-related Google search terms in total. The Google Trends data are then manually downloaded and consolidated into one single csv file. The EHR data are from athenahealth, a provider of cloud-based services and mobile applications for medical groups and health systems (www.athenahealth.com). The data we used are historical values of three nationally aggregated weekly fraction from total patient visit counts: (a) flu visit counts, (b) ILI visit counts, and (c) unspecified viral or ILI visit counts. That is, data reported are rounded fraction of each type of counts to total patient visit counts. The EHR data are available in real time starting from July 2009.