dismod iii

41
UNIVERSITY OF WASHINGTON DisMod III Abraham D. Flaxman JSM Vancouver, 2010 Integrated systems modeling for disease burden’s long tail

Upload: kane

Post on 24-Feb-2016

96 views

Category:

Documents


0 download

DESCRIPTION

DisMod III. Integrated systems modeling for disease burden’s long tail. Abraham D. Flaxman JSM Vancouver, 2010. Introduction. For Global Burden of Disease Study (GBD) : Must estimate incidence and duration for more than 250 diseases (by Nov 2010) - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: DisMod  III

UNIVERSITY OF WASHINGTON

DisMod III

Abraham D. Flaxman

JSM Vancouver, 2010

Integrated systems modeling for disease burden’s long tail

Page 2: DisMod  III

2

Introduction

For Global Burden of Disease Study (GBD) :• Must estimate incidence and duration for more than 250

diseases (by Nov 2010)• Estimates based on review of all available data, developed by

44 expert groups• Need estimates for 21 world regions, for males and females,

for 1990 and 2005 (and 2010)

How?

Page 3: DisMod  III

3

Introduction

For Global Burden of Disease Study (GBD) :• Must estimate incidence and duration for more than 250

diseases (by Nov 2010)• Estimates based on review of all available data, developed by

44 expert groups (these data are inconsistent) • Need estimates for 21 world regions, for males and females,

for 1990 and 2005 (and 2010)

How?

Page 4: DisMod  III

4

DisMod III Methods Outline

• Consistency of epidemiological parameters• Bayesian priors• Borrowing strength between regions• Web 2.0 interface

• Example Application - Guillain-Barré syndrome

Page 5: DisMod  III

Some Example Data - Dementia

Page 6: DisMod  III

Some Example Data - Anxiety

Page 7: DisMod  III

Compartmental Model for Consistency

7

Page 8: DisMod  III

8

Bayesian Statistical Model

Page 9: DisMod  III

9

Bayesian Statistical Model

Page 10: DisMod  III

10

Bayesian Inference via MCMC

• Computationally intensive, but possible

• Allows expert priors

Page 11: DisMod  III

DisMod Expert Priors

• Smoothing• Heterogeneity• Level bounds /

values• Increasing,

decreasing, unimodal

11

Page 12: DisMod  III

Expert Priors: Smoothing

12

Page 13: DisMod  III

Expert Priors: Smoothing

13

Page 14: DisMod  III

Expert Priors: Smoothing

14

Page 15: DisMod  III

Expert Priors: Smoothing

15

Page 16: DisMod  III

Expert Priors: Monotonicity

16

Page 17: DisMod  III

17

DisMod generates consistent estimates

Page 18: DisMod  III

18

DisMod generates consistent estimates

Page 20: DisMod  III

20

Statistical Model

Page 21: DisMod  III

DisMod Empirical Priors

21

Page 22: DisMod  III

DisMod Empirical Priors

22

Page 23: DisMod  III

DisMod Empirical Priors

23

Page 24: DisMod  III

DisMod Empirical Priors

24

Page 25: DisMod  III

Burden of Disease Workflow

Clean Data

• Check format• Check definitions with expert group• Check definitions in original data source• Clean as necessary

Load Data

• Explore in STATA or other general programs

• Explore in DisMod III• Incorporate additional data if necessary

Analyze Data

• Run Data• Adjust Expert Priors, adjust covariates• Discuss with Expert Groups• Repeat as necessary

Output Data

• Graphs, tables, STATA• Validation of results with other sources• Share results with expert groups

25

Page 26: DisMod  III

DisMod III

• Web-based User Interface

26

Page 27: DisMod  III

DisMod III

27

Page 28: DisMod  III

DisMod III

28

Page 29: DisMod  III

DisMod III

29

Page 30: DisMod  III

DisMod Disease View

Page 31: DisMod  III

DisMod Expert Priors

• Smoothing• Heterogeneity• Level bounds /

values• Increasing,

decreasing, unimodal

31

Page 32: DisMod  III

DisMod Covariates

32

Page 33: DisMod  III

DisMod Status Panel

33

Page 34: DisMod  III

34

Validation by Simulation Study

• Generate gold-standard data, 8400 rates with consistent incidence, prevalence, remission, excess-mortality

• Sample small portion of data, with noisy data generation model• Run DisMod III on the sample

Gold Standard Hold-out Cross-validationMedian Error Median Error Uncertainty

Interval CoverageAbsolute

(per 10,000)Relative Absolute

(per 10,000)Relative

Incidence 7.5 11 % 8.1 3.2 % 94 %

Prevalence 22 16 % 77 3.2 % 73 %

Remission 0.13 0.12

Excess Mortality

7.5 4.4 % 13 2.9 % 98 %

Duration 2.5 years 15%

Page 35: DisMod  III

DisMod Example: Guillain-Barré syndrome (GBS)

• Autoimmune disorder affecting the peripheral nervous system following an infectious disease

• Characterised by an ascending paralysis, spreading from legs to upper limbs and face

Page 36: DisMod  III

GBS data inputs

• Incidence• Remission• Mortality set to 0 after adjusting incidence by pooled case-

fatality assuming that disease specific mortality risk is early in disease with no further excess mortality thereafter

Page 37: DisMod  III

2005 GBS model posteriors

Page 38: DisMod  III

GBS Incidence in females, 1990

Page 39: DisMod  III

GBS Incidence in females, 2005

Page 40: DisMod  III

40

Conclusion and Lessons Learned

• Systematic literature review quality are crucialo Precious raw material that DisMod runs on…o …or GIGO?

• Expert knowledge from Doctors and Epidemiologists is crucialo Bayesian Priors will affect output, especially for parameters

without much datao Covariate selection will affect output, especially in regions without

much data

Page 41: DisMod  III

41

Acknowledgements

• DisMod Visionarieso Chris Murrayo Moshen Naghavio Theo Voso Rafael Lozanoo Steve Limo Colin Matherso Majid Ezzatio Jan Barendregto Rebecca Cooley

• DisMod Software Engineero Jiaji Du

• DisMod Early Adopterso Jed Baloreo Allyne Dellosantoso Samath Dharmaratneo Merhdad Forouzano Maya Mascarenhaso Nate Nairo Rosanna Normano Farshad Purmaleko Saied Shahrazo Gretchen Stevens