A Visualization Methodology for Studying Relations of Medical Data via Extended Dependency Networks

Measuring Similarity by Prediction Class between Biomedical Datasets via Fuzzy Unordered Rule Induction

“…The two datasets are subject to SPC measure by following the steps in Section 2. For the sake of objective comparison with existing methods, the following two methods are used -Pearson correlation as reported in [10] by using Dependency Network, and Mutual Information Score [15] which is a measure of mutual dependence of two sets of variables. The formula for Mutual Information as reported in [16] is defined as follow:…”

“…In this paper we present an alternative similarity measure called Similarity by Prediction Class (SPC). The method is extended from our previous work [10] called Dependency Network that displays out all the attributes and their respective predictive strengths to a disease, also inter-relations between symptoms across different diseases can be inferred. Using functions feature selection and information gain in inducing a predictive model, a Dependency Network is built by assigning the attributes of some disease significance values.…”

“…The underlying logic is a set of formula for quantitatively deriving a relational measure for this indirect dependency across diseases. Readers may find the mathematical definitions in [10]. One technical challenge in implementing a Dependency Network is the need of merging two or more medical datasets that have different dimensions in columns and rows though they may share some common attributes.…”