Sensitivity of Model-Based Epidemiological Parameter Estimation to Model Assumptions