Optimal feature with modified bi-directional long short-term memory for big data classification in healthcare application

Artificial intelligence together with its applications are advancing in all fields, particularly medical science. A considerable quantity of clinical data is available, yet the vast majority of it is wasted. It will be valuable in diagnosing human life early if it is applied effectively. An excellent classification system assists medical practitioners in providing a suitable diagnosis at a younger stage. Medical data typically contain a large number of features, and including all of them in decision making may result in over fitting of the classification model, lowering accuracy. As a result, an effective dimensionality reduction strategy that minimizes the amount of features while also improving classification accuracy is required. Therefore in this study, efficient dimension reduction-based medical data classification is proposed. The proposed approach is made up of three stages namely, pre-processing, feature selection, and classification. Initially, the medical data are collected and pre-processed. Then, to minimize computational complexity and time consumption, optimal features are selected using the adaptive rat swarm optimization (ARWO) algorithm. Then, the selected features are given to the modified bi-directional long short-term memory (MBi-LSTM) classifier to classify data as normal or abnormal. Then, the performance of the model is evaluated using different datasets and the effectiveness of the proposed model is analyzed in terms of different metrics. The proposed model implemented in python.