SEIR Model Based Epidemic Transmission Risk Deep Prediction

Abstract

Amidst ongoing efforts to manage the spread of infectious diseases, measures are taken to safeguard the resumption of on-campus activities. This ensures the academic continuity and the safety of the community. To achieve this, an analysis of disease transmission risks has been conducted, focusing on areas within campuses such as dining halls, lecture theaters, and classrooms. By leveraging an enhanced Susceptible-Exposed-Infectious-Removed (SEIR) model, we have developed a risk assessment model that takes into account dynamics between susceptible, latent, infected and displaced groups. After the extraction of relevant features, features undergo preprocessing steps. They are monotonically incremented and smoothed to eliminate noise, and then serve as input and labels for training stacked denoising autoencoder. The outcome of analysis indicates that the implementation of interventions can significantly mitigate the spread of disease. It can decrease the frequency of infection interactions, lower the transmission rate, and reduce the peak numbers of infected and latent cases by 61% and 72%, respectively. In essence, our approach has proven to be effective in controlling the spread of diseases in key university areas. Moreover, it provides an accurate predictive model for the number of infections, offering a valuable tool for managing and preventing outbreaks within these communities.