Modeling Cancer Mortality Risk Using Geostatistical Methods
Author Information
Author(s): Pierre Goovaerts
Primary Institution: BioMedware, Inc.
Hypothesis
Can geostatistical simulation improve the reliability of cancer mortality risk estimates from sparsely populated areas?
Conclusion
The study demonstrates that geostatistical simulation provides more realistic risk maps and allows for better visualization of uncertainty in cancer mortality data.
Supporting Evidence
- Geostatistical simulation generates risk maps that are more variable than smooth risk maps.
- Local cluster analysis shows lower reliability for pancreatic cancer classifications compared to breast cancer.
- 500 realizations are needed for stable classification of pancreatic cancer risks.
Takeaway
This study shows how scientists can create better maps to understand cancer risks in different areas, helping to find places that need more attention.
Methodology
The study used Poisson kriging and p-field simulation to generate multiple realizations of cancer mortality risk maps.
Potential Biases
Potential bias due to smoothing methods that may obscure local variations in risk.
Limitations
The study primarily focuses on white females and may not generalize to other demographics.
Participant Demographics
White females in 295 US counties from 1970 to 1994.
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website