Grant Number:	5R44CA065366-03 Interpret this number
Primary Investigator:	Jacquez, Geoffrey
Organization:	Biomedware
Project Title:	Software and Statistical Methods for Uncertain Locations
Fiscal Year:	1998

We propose to develop software can cluster statistics appropriate when the exact space-time location of health events are known. Health professionals are investigating an increasing number of possible disease clusters, and statistical tests play an important role in cluster description and analysis. Existing cluster statistics assume precise data, when in reality health events are often imprecise (e.g. place-of- residence is known only to the census district or zip-code) and uncertain (e.g. 'I first became ill sometime in 1985'). Most cluster statistics can be written as the cross product of two matrices where one matrix reflects nearest neighbor, distance of adjacency relationships and the second matrix is health related (e.g. case-control identities). This research will explore a general approach to clustering which incorporates uncertainty regarding space-time locations into the nearest neighbor, distance or adjacency relationship. Because the approach is general the proposed methods can be used with almost all exiting cluster tests. In phase 1 we will determine feasibility by implementing this general approach for Cuzick & Edwards (nearest neighbor-based), Mantel's (distance-based) and Knox's (adjacency-based) tests. The delivery of the prototype software and Manual at the end of phase 1 will be the criterion for demonstrating project feasibility. In phase 2 we will extend the approach to 10 other cluster tests and evaluate the fuzzy clustering algorithms using statistical power comparisons based on 3 realistic disease simulations. PROPOSED COMMERCIAL APPLICATION: The resulting software will be a powerful tool for the statistical description and detection of realistic clusters of health events characterized by uncertain space-time locations.





None