Using data mining techniques to identify factors associated with medication non-adherence

Abstract

Techniques for data mining and machine learning have recently become widely used in the healthcare industry. The goal of this research is to create an automated method of disease diagnosis. Here, three distinct methods for disease diagnosis utilizing data mining techniques are proposed: the hybrid fuzzy decision making tree approach, the association rule-based approach, and the efficient hybrid approach. The first strategy suggests an effective hybrid technique to lower the number of outliers. In data mining, outlier detection is a current study topic. The items that reside outside of the clusters are highlighted and identified as outliers if clustering techniques are applied. However, a small number of unidentified pieces might be added to the cluster. Therefore, it becomes vital to identify and remove such data that has been merged with the clusters in order to fully remove the unnecessary data from the dataset. The suggested method uses two datamining methods, Multilayer Neural Networks (MLN) and density-based K-means, to find outliers in a data set. When evaluated on the UCI repository, the suggested system outperforms the current ones in terms of disease prediction. The classification accuracy has increased while the time complexity has decreased.