Classification of Clinical data Abnormality using Neural Network and Machine Learning Models

Abstract

In the era of data explosion, the need of great importance is to recover data or information that is intriguing. This has to be mined from the raw facts available, for which various data mining techniques are employed. Clinical data are perceived to be a critical corporate resource and give essential proof of a medication's adequacy and its potential monetary incentive to the market. A regular structure of clinical data is an arrangement of perceptions of clinical parameters taken at various time minutes. Hence clinical evidence is time-variant. In this kind of contexts, the temporal dimension of information is a crucial variable that ought to be considered in the mining procedure. Using valid data mining methods to manage clinical data can enhance the speed and accuracy of the medical diagnosis and disease prediction process. Classification is a trivial data mining task that is being utilized by knowledge finding and decision support systems, especially in medical diagnosis and clinical decision support systems. In this process, we propose to develop a domain-specific intelligent classifier like neural networks, gradient boost methods to classify the heart abnormality of the patient. Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) used for attributes reduction in the dataset. To enhance the performance, we mentioned the classifier model tuned by a domain expert based on a dynamic error reporting scheme by further using Swarm Intelligence concept.