Try Free

Past Research

Geostatistical Software for the Boundary Analysis of Cancer Maps

NCI SBIR Phase I Completed research 2008-2008
Principal Investigator: Pierre Goovaerts, PhD

Summary

This SBIR project is developing the first GIS-based software to offer tools that are specifically designed for the boundary analysis of aggregated health data, providing: geostatistical modeling of the uncertainty attached to the spatial distribution of rates and its propagation through the computation of boundary statistics, interpolation of health outcomes and putative factors to the same spatial support to facilitate boundary overlap analysis, detection of significant boundaries accounting for the spatial pattern of rates and multiple testing correction, and visualization of changes in the location of these boundaries through time. This product will allow the detailed investigation of zones of abrupt changes in disease rates, and the exploration of relationships between health outcomes and potential factors, such as environmental exposures, socio-economic conditions, and cancer control methods, leading to: (1) a better understanding of the mechanisms/pathways/causes by which regions influence health outcomes, and (2) long-term quantification of the benefits of current strategies and policies for reducing the observed geographic disparities in cancer incidence, mortality and survival. Instructional materials will be developed to promote the use of this relatively new methodology among health scientists.

Phase I of the project will:

  • Develop and validate through simulation studies geostatistical techniques for modeling spatial uncertainty and statistical tests that account for spatial patterns and multiple testing in the detection of significant geographic boundaries in cancer rates and putative factors.
  • Conduct a requirements analysis to identify the optimal spatial methods and functionality to incorporate into the software: TerraSeer Space-Time Intelligence System™ (STIS™).
  • Develop and test a software prototype to detect, visualize and interpret significant boundaries in cancer maps.

This project takes advantage of significant opportunities for cross-disciplinary fertilization between a fundamental area of statistical science and application domains that stand to benefit tremendously from improved and more accessible statistical methods for description of spatial patterns, detection of significant geographic boundaries in cancer rates, and testing of relationships between boundaries for health outcomes and putative factors. By teaming international leaders in geostatistics, boundary analysis and disease mapping, this project will stimulate the development of theory and techniques in statistical analysis of cancer data, further improving our ability to interpret geographic and temporal variation in cancer incidence, and understand the causes underlying geographic disparities in incidence, mortality and survival rates.