Classical and Bayesian analyses of transmission experiments Jantien Backer and Thomas Hagenaars Epidemiology, Crisis management & Diagnostics Central Veterinary.

Slides:



Advertisements
Similar presentations
Introduction to Monte Carlo Markov chain (MCMC) methods
Advertisements

Stochastic spatio-temporal modelling methods in epidemiology and ecology Gavin J Gibson Heriot-Watt University NERC-EMS Workshop on Inference for Stochastic.
SAMPLE DESIGN: HOW MANY WILL BE IN THE SAMPLE—DESCRIPTIVE STUDIES ?
Study on HEV Vaccination Approach in Swine for Public Health protection Jeanette van der Goot, Jantien Backer, Leo van Leengoed, David Rodriquez Lazaro,
Uncertainty and confidence intervals Statistical estimation methods, Finse Friday , 12.45–14.05 Andreas Lindén.
Bayesian Estimation in MARK
Chapter 7 Title and Outline 1 7 Sampling Distributions and Point Estimation of Parameters 7-1 Point Estimation 7-2 Sampling Distributions and the Central.
Estimation  Samples are collected to estimate characteristics of the population of particular interest. Parameter – numerical characteristic of the population.
Fundamentals of Data Analysis Lecture 12 Methods of parametric estimation.
5 - 1 © 1997 Prentice-Hall, Inc. Importance of Normal Distribution n Describes many random processes or continuous phenomena n Can be used to approximate.
Sampling Distributions (§ )
BAYESIAN INFERENCE Sampling techniques
Bayesian inference Gil McVean, Department of Statistics Monday 17 th November 2008.
Statistical Inference Chapter 12/13. COMP 5340/6340 Statistical Inference2 Statistical Inference Given a sample of observations from a population, the.
Basics of Statistical Estimation. Learning Probabilities: Classical Approach Simplest case: Flipping a thumbtack tails heads True probability  is unknown.
Descriptive statistics Experiment  Data  Sample Statistics Experiment  Data  Sample Statistics Sample mean Sample mean Sample variance Sample variance.
Maximum Likelihood We have studied the OLS estimator. It only applies under certain assumptions In particular,  ~ N(0, 2 ) But what if the sampling distribution.
Using ranking and DCE data to value health states on the QALY scale using conventional and Bayesian methods Theresa Cain.
Lecture 7 1 Statistics Statistics: 1. Model 2. Estimation 3. Hypothesis test.
Standard error of estimate & Confidence interval.
Binary Variables (1) Coin flipping: heads=1, tails=0 Bernoulli Distribution.
Bayesian Inference, Basics Professor Wei Zhu 1. Bayes Theorem Bayesian statistics named after Thomas Bayes ( ) -- an English statistician, philosopher.
Basic Business Statistics, 11e © 2009 Prentice-Hall, Inc. Chap 8-1 Confidence Interval Estimation.
Prof. Dr. S. K. Bhattacharjee Department of Statistics University of Rajshahi.
PARAMETRIC STATISTICAL INFERENCE
A statistical model Μ is a set of distributions (or regression functions), e.g., all uni-modal, smooth distributions. Μ is called a parametric model if.
Bayesian inference review Objective –estimate unknown parameter  based on observations y. Result is given by probability distribution. Bayesian inference.
Exam I review Understanding the meaning of the terminology we use. Quick calculations that indicate understanding of the basis of methods. Many of the.
Lab3: Bayesian phylogenetic Inference and MCMC Department of Bioinformatics & Biostatistics, SJTU.
Likelihood Methods in Ecology November 16 th – 20 th, 2009 Millbrook, NY Instructors: Charles Canham and María Uriarte Teaching Assistant Liza Comita.
Determination of Sample Size: A Review of Statistical Theory
- 1 - Bayesian inference of binomial problem Estimating a probability from binomial data –Objective is to estimate unknown proportion (or probability of.
Statistical Inference for the Mean Objectives: (Chapter 9, DeCoursey) -To understand the terms: Null Hypothesis, Rejection Region, and Type I and II errors.
Week 41 Estimation – Posterior mean An alternative estimate to the posterior mode is the posterior mean. It is given by E(θ | s), whenever it exists. This.
: Chapter 3: Maximum-Likelihood and Baysian Parameter Estimation 1 Montri Karnjanadecha ac.th/~montri.
Dynamic Random Graph Modelling and Applications in the UK 2001 Foot-and-Mouth Epidemic Christopher G. Small Joint work with Yasaman Hosseinkashi, Shoja.
Bayesian Phylogenetics. Bayes Theorem Pr(Tree|Data) = Pr(Data|Tree) x Pr(Tree) Pr(Data)
Latent Class Regression Model Graphical Diagnostics Using an MCMC Estimation Procedure Elizabeth S. Garrett Scott L. Zeger Johns Hopkins University
On Predictive Modeling for Claim Severity Paper in Spring 2005 CAS Forum Glenn Meyers ISO Innovative Analytics Predictive Modeling Seminar September 19,
Summarizing Risk Analysis Results To quantify the risk of an output variable, 3 properties must be estimated: A measure of central tendency (e.g. µ ) A.
Inferences Concerning the Difference in Population Proportions (9.4) Previous sections (9.1,2,3): We compared the difference in the means (  1 -  2 )
M.Sc. in Economics Econometrics Module I Topic 4: Maximum Likelihood Estimation Carol Newman.
Stochastic Loss Reserving with the Collective Risk Model Glenn Meyers ISO Innovative Analytics Casualty Loss Reserving Seminar September 18, 2008.
Statistics Sampling Distributions and Point Estimation of Parameters Contents, figures, and exercises come from the textbook: Applied Statistics and Probability.
1 Chapter 8: Model Inference and Averaging Presented by Hui Fang.
Stochastic Frontier Models
10.1 – Estimating with Confidence. Recall: The Law of Large Numbers says the sample mean from a large SRS will be close to the unknown population mean.
Introduction: Metropolis-Hasting Sampler Purpose--To draw samples from a probability distribution There are three steps 1Propose a move from x to y 2Accept.
Parameter Estimation. Statistics Probability specified inferred Steam engine pump “prediction” “estimation”
Hypothesis Testing. Statistical Inference – dealing with parameter and model uncertainty  Confidence Intervals (credible intervals)  Hypothesis Tests.
Hierarchical Bayesian Analysis: Binomial Proportions Dwight Howard’s Game by Game Free Throw Success Rate – 2013/2014 NBA Season Data Source:
1 Spatial assessment of deprivation and mortality risk in Nova Scotia: Comparison between Bayesian and non-Bayesian approaches Prepared for 2008 CPHA Conference,
Statistical Inference for the Mean Objectives: (Chapter 8&9, DeCoursey) -To understand the terms variance and standard error of a sample mean, Null Hypothesis,
© 2001 Prentice-Hall, Inc.Chap 8-1 BA 201 Lecture 12 Confidence Interval Estimation.
Outline Historical note about Bayes’ rule Bayesian updating for probability density functions –Salary offer estimate Coin trials example Reading material:
Efficiency Measurement William Greene Stern School of Business New York University.
MCMC Stopping and Variance Estimation: Idea here is to first use multiple Chains from different initial conditions to determine a burn-in period so the.
Introducing Bayesian Approaches to Twin Data Analysis
Parameter Estimation 主講人:虞台文.
Bayesian Inference for Small Population Longevity Risk Modelling
More about Posterior Distributions
Bayesian Inference, Basics
Confidence Intervals Chapter 10 Section 1.
Ch13 Empirical Methods.
Determining Which Method to use
Sampling Distributions (§ )
Epidemiological parameters from transmission experiments: new methods for old data Simon Gubbins, David Schley & Ben Hu Transmission Biology Group The.
Applied Statistics and Probability for Engineers
How Confident Are You?.
Classical regression review
Presentation transcript:

Classical and Bayesian analyses of transmission experiments Jantien Backer and Thomas Hagenaars Epidemiology, Crisis management & Diagnostics Central Veterinary Institute of Wageningen UR The Netherlands InFER2011, 30 th of March 2011

2 Background Transmission experiments typical in veterinary epidemiology controlled environment known inoculation moments infection process monitored by regular sampling Analysis Maximum Likelihood Estimation: straightforward but discretizations and assumptions necessary Bayesian: more flexible (e.g. prior information, test characteristics) but more laborious Transmission experiments ideally suited for comparison of analyses

3 Outline Example transmission experiment MLE analysis Bayesian analysis Comparison MLE and Bayesian analyses simulated transmission experiments for low, medium and high R 0 how does ML estimate and median of posterior distribution relate? is the true value included in confidence and/or credible interval? Summary Next steps

4 Transmission experiment inoculated animal infectious animal contact (susceptible) animal removed animal day 0 day 1day day 21 vaccinated population of chickens challenged with Highly Pathogenic Avian Influenza H5N1 (data J.A. van der Goot)

† +++† † † † † † † † +++† ++++† +++† † † † † † Transmission experiment assumed: SIR model infection interval infectious interval removal interval

6 MLE analysis determine loglikelihood function maximize loglikelihood function MLE transmission rate parameter MLE infectious period distribution MLE reproduction number R 0 construct 95% confidence interval from likelihood profile using likelihood ratio test

7 MLE analysis s j : start of contact e1 j : start of infection interval e2 j : end infection interval c j : censoring infectious period (boolean) T j : infectious period = ½ (r1 j + r2 j ) - ½ (i1 j + i2 j ) probability of escaping infection β : transmission rate parameter N : total number of animals I(t) : number of infectious animals at time t μ : average infectious period σ : standard deviation infectious period

8 s j : start of contact e1 j : start of infection interval e2 j : end of infection interval c j : censoring infectious period (boolean) T j : infectious period = ½ (r1 j + r2 j ) - ½ (i1 j + i2 j ) β : transmission rate parameter N : total number of animals I(t) : number of infectious animals at time t μ : average infectious period σ : standard deviation infectious period probability of infection in interval (e1 j, e2 j ) MLE analysis

9 s j : start of contact e1 j : start of infection interval e2 j : end of infection interval c j : censoring infectious period (boolean) T j : infectious period = ½ (r1 j + r2 j ) - ½ (i1 j + i2 j ) β : transmission rate parameter N : total number of animals I(t) : number of infectious animals at time t μ : average infectious period σ : standard deviation infectious period

10 MLE analysis s j : start contact e1 j : start infection interval e2 j : end infection interval c j : censoring infectious period (boolean) T j : infectious period = ½ (r1 j + r2 j ) - ½ (i1 j + i2 j ) β : transmission rate parameter N : total number of animals I(t) : number of infectious animals at time t μ : average infectious period σ : standard deviation of infectious period pdf infectious period distribution

11 MLE analysis s j : start contact e1 j : start infection interval e2 j : end infection interval c j : censoring infectious period (boolean) T j : infectious period = ½ (r1 j + r2 j ) - ½ (i1 j + i2 j ) β : transmission rate parameter N : total number of animals I(t) : number of infectious animals at time t μ : average infectious period σ : standard deviation of infectious period cdf infectious period distribution s j : start contact e1 j : start infection interval e2 j : end infection interval c j : censoring infectious period (boolean) T j : infectious period = ½ (r1 j + r2 j ) - ½ (i1 j + i2 j )

12 MLE analysis s j : start contact e1 j : start infection interval e2 j : end infection interval c j : censoring infectious period (boolean) T j : infectious period = ½ (r1 j + r2 j ) - ½ (i1 j + i2 j ) β : transmission rate parameter N : total number of animals I(t) : number of infectious animals at time t μ : average infectious period σ : standard deviation of infectious period

13 MLE analysis β = 0.82 (0.41 – 1.46) day -1 μ = 8.5 (6.4 – 12.2) days σ = 5.6 (3.7 – 9.9) days R 0 = βμ = 7.0 (3.3 – 13.7)

14 Bayesian analysis determine likelihood function choose prior distributions uninformative Ga (0.01, 0.01) adjust proposal distributions during convergence to achieve acceptance rate of 40% - 60% MCMC chain (length 10000) update infection, infectious and removal moments: Metropolis-Hastings sampling ( normal proposal distributions) update β: Gibbs sampling update μ and σ: Metropolis-Hastings sampling (gamma proposal distributions) construct 95% credible interval from posterior parameter distributions

15 Bayesian analysis s j : start of contact e j : infection moment c j : censoring infectious period (boolean) T j : infectious period = (r j - i j ) β : transmission rate parameter N : total number of animals I(t) : number of infectious animals at time t μ : average infectious period σ : standard deviation of infectious period

16 Bayesian analysis medβ = 0.79 ( ) medμ = 8.7 (6.5 – 12.5) medσ = 5.9 (3.9 – 10.5)medR 0 = 6.8 (3.2 – 13.4) β = 0.82 ( ) μ = 8.5 (6.4 – 12.2) σ = 5.6 (3.7 – 9.9) R 0 = 7.0 (3.3 – 13.7) transmission parameter βaverage infectious period µ standard deviation σ of infectious period distributionreproduction number R 0

17 Comparison MLE and Bayesian analyses Simulated transmission experiments SIR model 5 inoculated animals with 5 contact animals, two replicates transmission rate parameter β = (0.125, 0.5, 2) day -1 average infectious period μ = 4 days standard deviation infectious period σ = 2√2 (shape parameter of 2) reproduction number R 0 = (0.5, 2, 8) sampling intervals of one day end of experiment at day 14 in total 100 simulated transmission experiments per scenario # contact infections R 0 = 0.5R 0 = 2R 0 = 8

18 Comparison MLE and Bayesian analyses 95% confidence interval ML estimate 95% credible interval median parameter value transmission parameter β MLE coverage: 94/100 Bayesian coverage: 91/100

19 Comparison MLE and Bayesian analyses, R 0 = 2 95% confidence interval ML estimate 95% credible interval median parameter value transmission parameter β 94/100 91/100 average infectious period 93/100 94/100 standard deviation infectious period distribution 95/100 97/100 reproduction number R 0 92/100

20 Comparison MLE and Bayesian analyses, R 0 = 8 95% confidence interval ML estimate 95% credible interval median parameter value transmission parameter β 78/100 75/100 average infectious period 91/100 standard deviation infectious period distribution 91/100 reproduction number R 0 80/100 77/100

21 Comparison MLE and Bayesian analyses, R 0 = % confidence interval ML estimate 95% credible interval median parameter value transmission parameter β 85/100 82/100 average infectious period 91/100 92/100 standard deviation infectious period distribution 88/100 89/100 reproduction number R 0 83/100

22 Summary Results MLE and Bayesian analyses maximum likelihood estimate similar to median value of posterior confidence interval comparable to credible interval inclusion of true value in confidence and credible intervals comparable

23 Next steps Bayesian analysis include latent period estimation implement test characteristics extend to larger groups with unobserved infections

24 Comparison MLE and Bayesian analyses: latent period 95% confidence interval ML estimate 95% credible interval median parameter value assumed SEIR model average latent period of 2 days (and shape parameter of 4) reproduction number R 0 = 2 average latent period of all infected animals (with informative gamma prior) reproduction number R 0

Thank you This study was funded by the Dutch Ministry of Economic Affairs, Agriculture and Innovation