# Bayesian calibration of simulation models: A tutorial and an Australian smoking behaviour model

**Authors:** Stephen Wade, Marianne F Weber, Peter Sarich, Pavla Vaneckova, Silvia Behar-Harpaz, Preson J Ngo, Sonya Cressman, Coral E Gartner, John M Murray, Tony A Blakely, Emily Banks, Martin C Tammemagi, Karen Canfell, Michael Caruana  
**arXiv:** [2202.02923](https://arxiv.org/abs/2202.02923)  
**Date:** 2022-02-07  
**Summary page:** https://www.summarizepaper.com/en/paper/2202.02923v1/

## Key points

- Calibration of simulation models using Bayesian statistics in various scientific disciplines
- Synthesis of multiple data sources and calculation of uncertainty in model output
- Common challenges and step-by-step process for successful calibration
- Analyses and approaches emerging from applied sciences
- Demonstration of steps and computations involved in calibration using a compartmental model of tobacco smoking behavior in Australia
- Findings on smoking behavior trends in Australia, including lowest proportion of individuals taking up smoking before age 20 since the early 20th century and highest quit rates
- Quantification of ex-smokers transitioning to "never smokers" using cross-sectional survey data
- Contribution to understanding through comprehensive tutorial and application to a specific model
- Potential for uncovering new insights from existing survey data using calibration

## Summary

This study focuses on the calibration of simulation models using Bayesian statistics in various scientific disciplines, including epidemiology, biology, ecology, and environmental science. The  allows for the synthesis of multiple data sources and the calculation of uncertainty in model output based on uncertainty in the calibration data. As more tutorials and studies are published, common challenges in  across these fields have become more apparent, leading to the emergence of a step-by-step process for successful calibration. The authors provide a statement outlining the key steps involved in  and discuss the analyses and approaches that have emerged from these applied sciences. This synthesis of  cuts across multiple scientific disciplines. To demonstrate these steps and provide further insight into the computations involved in , the researchers calibrated a compartmental model of tobacco smoking behavior in Australia. Their findings reveal that the proportion of individuals estimated to take up smoking before age 20 in 2016 was at its lowest value since the early 20th century, while quit rates were at their highest. Additionally, they quantified a novel outcome: the rate at which ex-smokers transitioned to reporting themselves as "never smokers" when surveyed later in life. This phenomenon had not been previously quantified using cross-sectional survey data. Overall, this study contributes to our understanding of  by providing a comprehensive tutorial and showcasing its application to a specific model. The findings shed light on tobacco smoking behavior trends in Australia and highlight the potential for utilizing  to uncover new insights from existing survey data.