Spatial Data Mining of a Population-Based Data Warehouse of Cancer in Mexico

In the health sciences area, data mining applications have had a fast growth due to its results concerning the generation of patterns of interest; however, its application to spatial population-based databases has been scant. This paper shows the results obtained by applying a spatial data mining system of our making to a real population-based data warehouse of cancer mortality in Mexico. The system consists of a pattern generator module, which uses a variant of a clustering algorithm proposed by us, and a spatial visualization module. Several interesting and potentially useful patterns of stomach cancer were found in the northwest of Mexico, which show promising results for extending the use of data mining in the area of epidemiology.