Optimal Bidirectional Long Short Term Memory Model for Medical Data Classification

Abstract

In recent times, medical field is being generated a large amount of data and it is hard to examine the particular features of the data. Generally, medical data classification is employed for the transformation of the description of medical diagnosis or processes to a standard statistical code called clinic coding. The recent development of artificial intelligence (AI) techniques paves a way for effective medical data classification. In this aspect, this paper designs a new rain optimization algorithm (ROA) based on bidirectional long short term memory (BiLSTM), called ROA-BiLSTM model for medical data classification. The ROA-BiLSTM model aims to determine the existence of the diseases from the available medical data. The ROA-BiLSTM model involves a 3-stage process namely preprocessing, classification, and hyperparameter optimization. In addition, the BiLSTM based classification process is performed in which the hyperparameters are optimally modified by the use of ROA and thereby boosts the overall performance. A wide range of simulations was carried out on the benchmark dataset and the performance of the ROA-BiLSTM model is investigated under different aspects. The experimental results highlighted the betterment of the ROA-BiLSTM model over the other compared methods.