Data-Driven Public Health Surveillance: Identifying Influenza Trends via Social

Abstract

The internet era makes it appear outdated to monitor and identify influenza using traditional methods. Among the long-term public health problems that influenza might exacerbate include diabetes, asthma, congestive heart failure, sinus infections, ear infections, and bacterial pneumonia. Deep learning (DL) techniques for influenza identification are more efficient than traditional approaches in terms of logistics and cost. The benefit of influenza prediction lies in its ability to minimize morbidity and mortality by allowing relevant departments to implement appropriate preventative and control actions after evaluating forecasted data. This research develops a Runge Kutta optimized Dynamic Gated recurrent unit (RKO-DGRU) public health with for influenza identification. Initially, the dataset is collected from kaggle and preprocessed utilizing the lemmatization method. Our approach can result in a sensitivity of 86.69%, specificity of 93.68%, and 97.5% accuracy. The findings highlight the possibility of applying DL approaches to efficiently identify and categorize influenza using data gleaned from conversations on open networks. It can thus provide efficient ways to stop and manage an Influenza epidemic.