Identification of cancer rules in Viet Nam by network modularity

Abstract

Data clustering tools can uncover new knowledge to be used in cancer diagnosis and treatment. In this study, we proposed a novel method to cluster records of a relation. First, we designed an algorithm that calculates the similarity between record pairs of the relation, and then this similarity measure was used to generate a network corresponding to the relation. Finally, we used a Network science technique to detect clusters of records from the network and extract insights from the clusters. Applying the method to mine a cancer-screening dataset at the Vietnam Central Cancer Hospital with over 177,000 records, we have discovered several new cancer laws in Viet Nam, which contribute to cancer detection and treatment support. It is disclosed from these cancer rules that some types of cancer run in certain family lines and living places in Viet Nam. Clustering a relation by Network science approach can be a good choice for mining large-scale relational data.