A Bayesian space-time model for clustering areal units based on their disease trends

Gary Napier, Duncan Lee, Chris Robertson, Andrew Lawson

Research output: Contribution to journalArticle

Abstract

Population-level disease risk across a set of non-overlapping areal units varies in space and time, and a large research literature has developed methodology for identifying clusters of areal units exhibiting elevated risks. However, almost no research has extended the clustering paradigm to identify groups of areal units exhibiting similar temporal disease trends. We present a novel Bayesian hierarchical mixture model for achieving this goal, with inference based on a Metropolis coupled Markov chain Monte Carlo ((MC)3) algorithm. The effectiveness of the (MC)3 algorithm compared to a standard MCMC implementation is demonstrated in a simulation study, and the methodology is motivated by two important case studies in Scotland. The first concerns the impact on measles susceptibility of the discredited paper linking the Measles, Mumps and Rubella vaccination to an increased risk of Autism, and investigates whether all areas in Scotland were equally affected. The second concerns respiratory hospitalisations, and investigates over a 10 year period which parts of Glasgow have shown increased, decreased, and no change in risk.
LanguageEnglish
Number of pages17
JournalBiostatistics
Early online date18 Jun 2018
DOIs
StateE-pub ahead of print - 18 Jun 2018

Fingerprint

Space-time Models
Clustering
Unit
Markov Chain Monte Carlo
Group of Units
Methodology
Vaccination
Hierarchical Model
Mixture Model
Susceptibility
Linking
Paradigm
Simulation Study
Vary
Trends
Markov chain Monte Carlo
Scotland

Keywords

  • health inequalities
  • metropolis coupled Markov chain Monte Carlo ((MC)3 ) simulation
  • space-time disease mapping
  • trend estimation

Cite this

@article{3e508ed718344460a5b80a8d35a88d2f,
title = "A Bayesian space-time model for clustering areal units based on their disease trends",
abstract = "Population-level disease risk across a set of non-overlapping areal units varies in space and time, and a large research literature has developed methodology for identifying clusters of areal units exhibiting elevated risks. However, almost no research has extended the clustering paradigm to identify groups of areal units exhibiting similar temporal disease trends. We present a novel Bayesian hierarchical mixture model for achieving this goal, with inference based on a Metropolis coupled Markov chain Monte Carlo ((MC)3) algorithm. The effectiveness of the (MC)3 algorithm compared to a standard MCMC implementation is demonstrated in a simulation study, and the methodology is motivated by two important case studies in Scotland. The first concerns the impact on measles susceptibility of the discredited paper linking the Measles, Mumps and Rubella vaccination to an increased risk of Autism, and investigates whether all areas in Scotland were equally affected. The second concerns respiratory hospitalisations, and investigates over a 10 year period which parts of Glasgow have shown increased, decreased, and no change in risk.",
keywords = "health inequalities, metropolis coupled Markov chain Monte Carlo ((MC)3 ) simulation, space-time disease mapping, trend estimation",
author = "Gary Napier and Duncan Lee and Chris Robertson and Andrew Lawson",
year = "2018",
month = "6",
day = "18",
doi = "10.1093/biostatistics/kxy024",
language = "English",
journal = "Biostatistics",
issn = "1465-4644",

}

A Bayesian space-time model for clustering areal units based on their disease trends. / Napier, Gary; Lee, Duncan; Robertson, Chris; Lawson, Andrew.

In: Biostatistics, 18.06.2018.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A Bayesian space-time model for clustering areal units based on their disease trends

AU - Napier,Gary

AU - Lee,Duncan

AU - Robertson,Chris

AU - Lawson,Andrew

PY - 2018/6/18

Y1 - 2018/6/18

N2 - Population-level disease risk across a set of non-overlapping areal units varies in space and time, and a large research literature has developed methodology for identifying clusters of areal units exhibiting elevated risks. However, almost no research has extended the clustering paradigm to identify groups of areal units exhibiting similar temporal disease trends. We present a novel Bayesian hierarchical mixture model for achieving this goal, with inference based on a Metropolis coupled Markov chain Monte Carlo ((MC)3) algorithm. The effectiveness of the (MC)3 algorithm compared to a standard MCMC implementation is demonstrated in a simulation study, and the methodology is motivated by two important case studies in Scotland. The first concerns the impact on measles susceptibility of the discredited paper linking the Measles, Mumps and Rubella vaccination to an increased risk of Autism, and investigates whether all areas in Scotland were equally affected. The second concerns respiratory hospitalisations, and investigates over a 10 year period which parts of Glasgow have shown increased, decreased, and no change in risk.

AB - Population-level disease risk across a set of non-overlapping areal units varies in space and time, and a large research literature has developed methodology for identifying clusters of areal units exhibiting elevated risks. However, almost no research has extended the clustering paradigm to identify groups of areal units exhibiting similar temporal disease trends. We present a novel Bayesian hierarchical mixture model for achieving this goal, with inference based on a Metropolis coupled Markov chain Monte Carlo ((MC)3) algorithm. The effectiveness of the (MC)3 algorithm compared to a standard MCMC implementation is demonstrated in a simulation study, and the methodology is motivated by two important case studies in Scotland. The first concerns the impact on measles susceptibility of the discredited paper linking the Measles, Mumps and Rubella vaccination to an increased risk of Autism, and investigates whether all areas in Scotland were equally affected. The second concerns respiratory hospitalisations, and investigates over a 10 year period which parts of Glasgow have shown increased, decreased, and no change in risk.

KW - health inequalities

KW - metropolis coupled Markov chain Monte Carlo ((MC)3 ) simulation

KW - space-time disease mapping

KW - trend estimation

UR - https://academic.oup.com/biostatistics

U2 - 10.1093/biostatistics/kxy024

DO - 10.1093/biostatistics/kxy024

M3 - Article

JO - Biostatistics

T2 - Biostatistics

JF - Biostatistics

SN - 1465-4644

ER -