Microsimulation model calibration using incremental mixture approximate Bayesian computation

DOI10.1214/19-AOAS1279zbMATH Open1435.62398arXiv1804.02090MaRDI QIDQ128019FDOQ128019

Authors: Carolyn Rutter, Jonathan Ozik, Maria DeYoreo, Nicholson T Collier, Carolyn M. Rutter, Jonathan Ozik, Nicholson Collier, Maria DeYoreo 

Publication date: 6 April 2018

Published in: The Annals of Applied Statistics (Search for Journal in Brave)

Abstract: Microsimulation models (MSMs) are used to predict population-level effects of health care policies by simulating individual-level outcomes. Simulated outcomes are governed by unknown parameters that are chosen so that the model accurately predicts specific targets, a process referred to as model calibration. Calibration targets can come from randomized controlled trials, observational studies, and expert opinion, and are typically summary statistics. A well calibrated model can reproduce a wide range of targets. MSM calibration generally involves searching a high dimensional parameter space and predicting many targets through model simulation. This requires efficient methods for exploring the parameter space and sufficient computational resources. We develop Incremental Mixture Approximate Bayesian Computation (IMABC) as a method for MSM calibration and implement it via a high-performance computing workflow, which provides the necessary computational scale. IMABC begins with a rejection-based approximate Bayesian computation (ABC) step, drawing a sample of parameters from the prior distribution and simulating calibration targets. Next, the sample is iteratively updated by drawing additional points from a mixture of multivariate normal distributions, centered at the points that yield simulated targets that are near observed targets. Posterior estimates are obtained by weighting sampled parameter vectors to account for the adaptive sampling scheme. We demonstrate IMABC by calibrating a MSM for the natural history of colorectal cancer to obtain simulated draws from the joint posterior distribution of model parameters.

Full work available at URL: https://arxiv.org/abs/1804.02090

Recommendations

• Bayesian calibration of microsimulation models
• Statistical inference in micro-simulation models: Incorporating external information
• Simulation based calibration using extended balanced augmented empirical likelihood
• Bayesian calibration of multistate stochastic simulators
• Recalibration: a post-processing method for approximate Bayesian computation

zbMATH Keywords

adaptive ABCagent-based modelscolorectal cancerincremental mixture approximate Bayesian computation (IMABC)microsimulation models (MSMs)

Mathematics Subject Classification ID

Bayesian inference (62F15) Applications of statistics to biology and medical sciences; meta analysis (62P10)

Cites Work

• A Simplex Method for Function Minimization
• Sequential Monte Carlo without likelihoods
• Monte Carlo strategies in scientific computing
• Approximate Bayesian computational methods
• A unified approach to the Richards-model family for use in growth analyses: why we need only two model forms
• Bayesian calibration of microsimulation models
• Estimating and Projecting Trends in HIV/AIDS Generalized Epidemics Using Incremental Mixture Importance Sampling
• Adaptive approximate Bayesian computation
• Asymptotic properties of approximate Bayesian computation
• Handbook of Approximate Bayesian Computation
• On the asymptotic efficiency of approximate Bayesian computation estimators
• Title not available (Why is that?)
• Approximate Bayesian computation and simulation-based inference for complex stochastic epidemic models

Cited In (8)

• Analyzing stochastic computer models: a review with opportunities
• Statistical inference in micro-simulation models: Incorporating external information
• Bayesian calibration of microsimulation models
• A review on computer model calibration
• Batch-sequential design and heteroskedastic surrogate modeling for delta smelt conservation
• Bayesian inference using Gaussian process surrogates in cancer modeling
• Bayesian hidden Markov model for natural history of colorectal cancer: handling misclassified observations, varying observation schemes and unobserved data
• imabc

Uses Software

• R

This page was built for publication: Microsimulation model calibration using incremental mixture approximate Bayesian computation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q128019)