cancer etiology and natural history: a web tool for age ... › apc › documentation...cancer...
TRANSCRIPT
![Page 1: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/1.jpg)
Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis
Philip S. Rosenberg David P. Check
William F. Anderson &
The NCI Center for Biomedical Informatics and Information Technology (CBIIT)
Biostatistics Branch
Division of Cancer Epidemiology and Genetics National Cancer Institute
06 March 2014, DCEG Seminar 1
![Page 2: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/2.jpg)
2
![Page 3: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/3.jpg)
• What does it do? • Why is that important? • Who built it, and how? • How do I use it?
3
![Page 4: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/4.jpg)
• Robert Shirley, NCI CBIIT • Sue Pan, NCI CBIIT • Larry Brem, Leidos Biomedical Research, Inc. (NCI CBIIT Dev Team Contractor) • Brent Coffey, Leidos Biomedical Research, Inc. (NCI CBIIT Dev Team Contractor) • Shaun Einolf, Leidos Biomedical Research, Inc. (NCI CBIIT Dev Team Contractor) • Sula Rajapakse, Leidos Biomedical Research, Inc. (NCI CBIIT Dev Team Contractor • Cuong Nguyen, SRA International, Inc. (NCI CBIIT Systems Team Contractor)
• Carl McCabe, Office of Division Operations and Analysis • Sholom Wacholder, Nicolas Wentzensen, Christine Fermo
• http://analysistools.nci.nih.gov/meanstorisk/
Acknowledgments
4
![Page 5: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/5.jpg)
Outline
• The APC Model • Overview • Examples from the literature
• The Web Tool • What’s in it (and why) • How it works
5
![Page 6: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/6.jpg)
Overview Examples from the literature
• Macro-epidemiological model for population-based cancer surveillance data
o Incidence and Mortality o SEER, IARC, other large-scale open cohorts
• Parametric approach o complements traditional descriptive approaches
• Quantification (via parameters and functions) o Burden o Trends o Natural History o Etiology o Disparity
APC Model: Overview
6
![Page 7: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/7.jpg)
Overview Examples from the literature
Rate matrix or Lexis diagram for invasive female breast cancer.
Rosenberg P S , and Anderson W F Cancer Epidemiol Biomarkers Prev 2011;20:1263-1268
©2011 by American Association for Cancer Research
A registry is a cohort of cohorts . . . APC Model: Data
We can study patterns both cross-sectionally (over time) and longitudinally (by cohort).
7
![Page 8: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/8.jpg)
ρac = µ + αL + π L( ) a − a ( )+ π L + γ L( ) c − c ( )+ ˜ α a + ˜ π p + ˜ γ c
Longitudinal Age Trend Net Drift Deviations
ρap = µ + αL − γ L( ) a − a ( )+ π L + γ L( ) p − p ( )+ ˜ α a + ˜ π p + ˜ γ c
Cross-Sectional Age Trend
Overview Examples from the literature
APC Model: Parameters from Data Longitudinal Form
Cross-sectional Form
8
![Page 9: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/9.jpg)
Overview Examples from the literature
Through independent and collaborative descriptive studies, we developed a panel of standard and novel functions** and corresponding hypothesis tests that appear to be effective in identifying signatures or patterns in disease rates for many types of cancers.
APC Model: Putting the pieces together
** linear combination of estimable parameters in the APC Model
9
![Page 10: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/10.jpg)
•Net Drift is the single most important parameter! o Model analogue of EAPC adjusted for cohort effects o Determines ratio of Longitudinal to Cross-sectional Age
Curves
•Age effects (Longitudinal and Cross-sectional Age Curves) •Period effects (Fitted Temporal Trends, Period RR) •Cohort effects (Cohort RR; Local Drifts = age-specific EAPC)
•The Significance Test for Local Drifts is the second most important APC statistic!
o Tells you if you have important cohort effects
Overview Examples from the literature
APC Model: Key Parameters, Functions, and Tests
10
![Page 11: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/11.jpg)
Outline
• The APC Model • Overview • Examples from the literature
• The Web Tool • What’s in it (and why) • How it works
Overview Examples from the literature
11
![Page 12: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/12.jpg)
Overview Examples from the literature
Age Effects (Longitudinal Age Curve)
Cross-Sectional
Longitudinal
12
![Page 13: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/13.jpg)
Overview Examples from the literature
Period Effects (Deviations, Period RR)
13
![Page 14: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/14.jpg)
Overview Examples from the literature
Cohort Effects (Cohort RR)
California New York Alabama
14
![Page 15: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/15.jpg)
Overview Examples from the literature
Cohort Effects (Local Drifts)
Age-specific temporal
trends
Local Drifts
15
![Page 16: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/16.jpg)
Outline
• The APC Model • Overview • Examples from the literature
• The Web Tool • What’s in it (and why) • How it works
What’s in it (and why) How it works
16
![Page 17: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/17.jpg)
All of the APC functions (and corresponding hypothesis tests) shown in Examples from the literature are produced by the Web Tool.
What’s in it (and why) How it works
17
![Page 18: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/18.jpg)
What’s in it (and why) How it works
Key Functions
18
![Page 19: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/19.jpg)
What’s in it (and why) How it works
Hypothesis Tests
19
![Page 20: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/20.jpg)
• We paid a lot of attention to workflow.
• The Web Tool promotes reproducible research.
• We think it is really simple to use.
What’s in it (and why) How it works
Web Tool: Usability
20
![Page 21: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/21.jpg)
User (Browser)
Python + JavaScript (Front-End
Server)
R
(Back-End Server)
Web Tool: Architecture
MATLAB
(Prototyping)
SEERSTAT ↓ Excel csv
R Excel Text
What’s in it (and why) How it works
21
![Page 22: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/22.jpg)
What’s in it (and why) How it works
Data Input
22
![Page 23: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/23.jpg)
Data Input Using Excel What’s in it (and why) How it works
23
![Page 24: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/24.jpg)
Data Input Using CSV What’s in it (and why) How it works
24
![Page 25: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/25.jpg)
Click on this button . . .
What’s in it (and why) How it works
25
![Page 26: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/26.jpg)
Scroll down . . .
What’s in it (and why) How it works
26
![Page 27: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/27.jpg)
Hypothesis Tests . . .
What’s in it (and why) How it works
27
![Page 28: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/28.jpg)
What’s in it (and why) How it works
28
![Page 29: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/29.jpg)
Click on this button . . .
What’s in it (and why) How it works
29
![Page 30: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/30.jpg)
What’s in it (and why) How it works
Model Outputs in Excel
30
![Page 31: Cancer Etiology and Natural History: A Web Tool for Age ... › apc › documentation...Cancer Etiology and Natural History: A Web Tool for Age-Period-Cohort Analysis . Philip S. Rosenberg](https://reader033.vdocuments.net/reader033/viewer/2022042400/5f0e74f57e708231d43f5690/html5/thumbnails/31.jpg)
• What does it do? The tool fits the APC Model and serves up Model Outputs.
• Why is that important? Many cancers present complicated patterns. The outputs complement and extend standard descriptive methods.
• Who built it, and how? BB – concept, design, computations CBIIT – “Webification”
• How do I use it? http://analysistools.nci.nih.gov/apc/
Conclusions
31