Modelling under-reporting in epidemics

Area of Science:

• Epidemiology
• Mathematical Biology
• Biostatistics

Background:

• Under-reporting of infectious cases introduces significant bias in epidemic modeling.
• Accurate parameter estimation is vital for understanding disease dynamics and implementing control strategies.

Purpose of the Study:

• To investigate the impact of ignoring under-reporting on epidemic model parameter estimation.
• To develop and evaluate models that account for under-reporting in stochastic SIR (Susceptible-Infected-Recovered) epidemic models.
• To assess the estimability of reporting rates and other epidemic parameters under various reporting scenarios.

Main Methods:

• Stochastic Markovian SIR epidemic modeling incorporating diverse reporting processes.
• Bayesian inference framework utilizing data augmentation and reversible jump Markov chain Monte Carlo (MCMC) techniques.
• Analysis of constant, time-dependent, and source-dependent reporting probabilities.

Main Results:

• Ignoring under-reporting leads to under-estimation of the infection rate and reproduction number.
• Models accounting for under-reporting allow for more accurate estimation of epidemic parameters.
• The Bayesian approach with MCMC effectively handles incomplete data and complex reporting mechanisms.

Conclusions:

• Under-reporting significantly biases epidemic parameter estimation, necessitating its explicit inclusion in models.
• The developed Bayesian framework provides a robust method for estimating epidemic parameters even with incomplete reporting.
• Accurate estimation of reporting rates is feasible and crucial for reliable epidemic forecasting and management.