Computational and data sciences for health-GIS

Abstract

Computational and data sciences are transforming the entire science enterprise. In the arena of GIS, this is represented by the emergence of cyberGIS. We provide an overview of applying the cyberGIS approach to spatial analysis for health studies. We emphasize that cyberGIS is not just a service to traditional spatial analyses, but itself is an alternative approach to problem solving. Some fundamental and profound distinctions of cyberGIS approaches in health-GIS include the following: (1) they may greatly reduce the reliance on models or assumptions, and instead seek actual empirical evidence through mining a large amount of data or virtual empirical evidence generated through computation; (2) they tend to be non-parametric and tend to generate local solution; (3) they are scalable to high-resolution and less aggregated data; (4) they tend to be stochastic rather than deterministic; and (5) with these approaches, the large amount of data may not be only from input data-sets, but also from analytical workflows. We described the kernel ratio estimation for local intensity estimation, the restricted and controlled Monte Carlo for data disaggregation, and unrestricted and controlled Monte Carlo for statistical significance evaluation as examples of the cyberGIS approaches in health-GIS.