Model distinguishability and inference robustness in mechanisms of cholera transmission and loss of immunity

TL;DR

This study evaluates how different cholera transmission models can be distinguished based on data fit and how robust their inferences are, highlighting the importance of data quality and model structure in epidemic prediction.

Contribution

It systematically assesses model distinguishability and inference robustness in cholera modeling, demonstrating the limits of mechanistic inference from epidemic data alone.

Findings

All models fit data well, even if misspecified.

Model fit improves when the model's immunity loss matches the data-generating process.

Forecast accuracy improves after the epidemic peak.

Abstract

Mathematical models of cholera and waterborne disease vary widely in their structures, in terms of transmission pathways, loss of immunity, and other features. These differences may yield different predictions and parameter estimates from the same data. Given the increasing use of models to inform public health decision-making, it is important to assess distinguishability (whether models can be distinguished based on fit to data) and inference robustness (whether model inferences are robust to realistic variations in model structure). We examined the effects of uncertainty in model structure in epidemic cholera, testing a range of models based on known features of cholera epidemiology. We fit to simulated epidemic and long-term data, as well as data from the 2006 Angola epidemic. We evaluated model distinguishability based on data fit, and whether parameter values and forecasts can accurately be inferred from incidence data. In general, all models were able to successfully fit to all data sets, even if misspecified. However, in the long-term data, the best model fits were achieved when the loss of immunity form matched those of the model that simulated the data. Two transmission and reporting parameters were accurately estimated across all models, while the remaining showed broad variation across the different models and data sets. Forecasting efforts were not successful early, but once the epidemic peak had been achieved, most models showed similar accuracy. Our results suggest that we are unlikely to be able to infer mechanistic details from epidemic case data alone, underscoring the need for broader data collection. Nonetheless, with sufficient data, conclusions from forecasting and some parameter estimates were robust to variations in the model structure, and comparative modeling can help determine how variations in model structure affect conclusions drawn from models and data.