A hybrid clustering and classification model for chemical code based medical disease prediction

Abstract

As the number of biomedical documents sets and medical datasets are increasing in size and dimensions, finding
an essential key ICD based disease terms are difficultto extract in large training databases. Most of the
traditional approaches use static ICD code extraction for the medical disease classification process. In this paper,
a hybrid ICD-Disease clustering-based classification approach is designed and implemented on the large
databases. In this work, a hybrid graph-based clustering algorithm is implemented in order to optimize the data
clustering operation for the classification problem. Finally, a weighted neural network is applied on the clustered
features for classification process. Experimental results show that the present model has high computational
efficiency than the conventional models.