A Knowledge Driven Model: Extract Knowledge from High Dimensional Medical Databases

The extensive amount of spatial databases accumulated from various computed technology requires automated data mining tools to discover hidden and novel information from large and complex databases. In other words, the major concern is high dimensionality and complexity of spatial data which has created serious concerns among the researchers to retrieve effective and efficient clusters from large and complex spatial features. In this paper we have proposed a spatial clustering algorithm (SPAS) and Knowledge driven framework to discover clusters of variant shapes and size with domain specific knowledge. The application of our proposed algorithm is tested on real world spatial medical databases collected from SEER datasets which has record of Lung cancer patients from the year 1975 - 2008. The case includes information on patient´s gender, ZIP code of a patient´s residence, year of diagnosis, primary site, stage at diagnosis, and age group. Each record represents a diagnosed cancer case assigned to the patient´s residence at time of diagnosis. The objective of study was to discover effective and efficient spatial clusters with domain specific knowledge for futuristic decision making.