Modeling and forecasting infectious diseases

Abstract

In recent years, the modeling and forecasting of infectious diseases have received heightened attention, driven by the emergence and global spread of high-profile outbreaks such as the 2014 Ebola epidemic in West Africa and the COVID-19 pandemic. These health crises have highlighted the critical importance of having timely, accurate information about disease transmission dynamics to guide public health decisions and interventions effectively. The role of infectious disease modeling and forecasting extends beyond mere academic inquiry; it serves as a vital tool for public health officials, offering the potential to predict outbreaks, strategize interventions, allocate resources judiciously, and communicate risks to communities. Therefore, this dissertation endeavors to develop and refine systems capable of modeling epidemic dynamics and delivering precise forecasts for a variety of infectious diseases, thereby supporting enhanced public health preparedness and response. Firstly, we introduce a short-term forecasting system for COVID-19, employing the SEIR compartmental model to simulate the disease’s transmission and employing the ensemble Kalman filter (EnKF) to assimilate daily confirmed case data. This system generates three-day forecasts, updated daily as new data becomes available, with forecast accuracy assessed through absolute relative error metrics. Secondly, our work expands into the vector-borne disease – dengue fever, developing a novel forecasting model that integrates the SEIR-SEI disease model with the EnKF method. Initial twin experiments assimilated synthetic weekly dengue case data, leading to accurate state and parameter estimations and predictions. Subsequently, this model was applied to historical dengue outbreak data from Kaohsiung, Taiwan; Singapore; and Rio de Janeiro, Brazil, yielding forecasts that closely aligned with observed cases and dynamically estimating exposed and infectious human populations. A comparative accuracy assessment of this model against the SEIR-EnKF and SIR-EnKF models highlighted its superior forecasting capabilities. Third, we detail a real-time prediction system for West Nile Virus (WNV) that incorporates an adapted compartment model to account for the transmission dynamics among birds, mosquitoes, and humans, including asymptomatic cases and the influence of climatic factors. Using data assimilation techniques, we generate weekly WNV case forecasts for Colorado in 2023, providing valuable insights for public health planning. Comparative analyses underscore the enhanced forecast accuracy achieved by integrating climatic variables into our models. Fourth, we explore the transmission dynamics of Foot-and-Mouth Disease (FMD) among beef-cattle farms in southwest Kansas through a three-layer network model encompassing direct contacts (cattle movements), indirect contacts (truck visits), and information sharing. This study assesses the impact of these layers on FMD transmission, the effectiveness of information-sharing strategies in mitigating outbreaks, and the economic implications of infection and quarantine measures. Finally, our work on the development of the PICTUREE—Aedes web application represents a significant advancement in the monitoring and forecasting of dengue fever. By aggregating and visualizing dengue-related data from diverse sources, including temperature, precipitation, mosquito occurrences, and case reports, the application facilitates global risk assessments and forecasts of ongoing outbreaks. In conclusion, this dissertation presents a comprehensive suite of modeling and forecasting tools for various infectious diseases, offering significant potential to optimize public health responses through improved predictive accuracy and timely information dissemination.