Try Free

Past Research

Simulation Algorithms for Spatial Pattern Recognition

NCI SBIR II Completed research 2005-2006
Principal Investigator: Pierre Goovaerts, PhD

Summary

This SBIR project is developing neutral spatial models for use in statistical pattern recognition applied to vector (e.g. geographically aggregated cancer data) and raster (e.g. large grid of modeled exposure data or imagery of habitats suitable for development of vector-borne disease) datasets. We will develop fast simulation algorithms that condition on distributions of the data, and that incorporate spatial and multivariate dependencies. The specific aims of this project are to:

  • Conduct a requirements analysis to specify the neutral models and functionality to incorporate in the software.
  • Develop and test a software prototype to evaluate feasibility of the proposed models.
  • Propose a topology of neutral models and develop strategies to generate them and to conduct sensitivity analysis for investigating the impact of implicit assumptions (i.e. spatial autocorrelation or non-uniform risk) and number of realizations on test results.
  • Incorporate the neutral models in a commercially established software package.
  • Apply the software and methods to demonstrate the approach and its unique benefits for exposure and health risk assessment.

The Phase I research addressed the first two aims. It implemented three algorithms to generate neutral models and assessed the feasibility of using them in commercial software.  This Phase II project will accomplish aims three through five. Capitalizing on recent research by the PI, a topology of neutral models will be developed and the palette of available models will be used to assess the sensitivity of results to specification of the null hypothesis. The resulting set of neutral models will be programmed, documented, and applied in an existing commercial software package for exposure and health risk assessment. Last, this new methodology will be used to revisit the findings of a study conducted at Biomedware about cancer and exposure data in Long Island, New York (Jacquez and Greiling, 2003a,b).