Using Electronic Health Records for Influenza Surveillance and Prediction

  • Tseng, Yi-Ju (PI)
  • Huang, Ching Tai (CoPI)

Project: National Science and Technology CouncilNational Science and Technology Council Academic Grants

Project Details


Influenza is a contagious respiratory illness caused by influenza viruses. While influenza viruses can be detected throughout the year, they are most common during the fall and winter. Unlike common cold, influenza can cause serious illnesses and be life-threatening. Disease detection and surveillance provide epidemic knowledge that allows the Centers for Infectious Disease Control (CDC) and hospitals to deploy preventive measures and make optimal decisions. Monitoring, early detection, and prediction of influenza outbreaks can quicken life-saving public heath responses. CDC in Taiwan uses the information from reliable infectious disease surveillances to monitor the occurrence and spread of severe complicated influenza cases. Other data sources such as laboratory-based surveillance systems and the Real-time Outbreak and Disease Surveillance (RODS) system are used as additional sources for influenza surveillance. However, ensuring completeness and timeliness of all the data sources can be difficult because they rely on an extensive network of health professionals and laboratories; therefore, the reporting process could be different for various infectious diseases and among physicians and geographic areas. We aimed to define influenza and its severe complicated cases and to predict influenza activity using electronic medical records from island-wide hospitals belonging to a medical foundation in Taiwan, having 9,000 beds and serving 28,000 out-patients a day. The first step is to collect data regarding infectious disease from electronic medical records. Patients who had at least one influenza diagnosis code in 2007–2016 are initially included. We then apply data mining algorithms, including the autoregressive integrated moving average, support vector machine regression, and random forest regression, to predict the influenza activity. The proposed models can be used to support current influenza surveillances, thus providing an accurate, real-time estimate of influenza activity. We also expect that the influenza activity predictive model can be used as a decision support tool for improving influenza surveillance, controlling infection, and managing resources.

Project IDs

Project ID:PB10703-1480
External Project ID:MOST106-2221-E182-072
Effective start/end date01/08/1730/04/18


  • Influenza
  • Infectious Disease Surveillance
  • Electronic Medical Record
  • Data Mining
  • Predictive


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.