A novel optimization technique using deep learning approach for prediction of multiple sclerosis by implementing the architect of Hybrid—RNN & LSTM

Multiple sclerosis (MS) is a chronic autoimmune disease that affects the Central nervous system (CNS). MS is a neurological disease in which immune system gets affected. MS affects the human brain and the spinal cord due to loss of protective covering fluid. The disease rate gradually gets increased and World health organization has predicted that by the year of 2028 the rate of MS will raise up to 1,306.2 billion in India. An action has to been taken to reduce the cause of MS. A Machine learning (ML) technique has been introduced to predict the disease and increase its accuracy. An existing ML algorithm such as Random Forest (RF), Catboost, K-Nearest Neighbor (K-NN) and Naïve Bayes were implemented and tested the accuracy rate which leads to an average of about 82%. To enhance the MS prediction an advanced deep learning algorithm which includes the combination of Recurrent Neural Network (RNN) and Long Short-term memory (LSTM) were used in this research. As a result, these proposed algorithm uses dataset which is tested from Mexican patients having 19 attributes in stage 1 MS between the year 2006 to 2010. By using the RNN and LSTM, the earlier prediction of MS leads 90% of accuracy. In this proposed method the balanced dataset is used. Hence the specificity and sensitivity acquired for the proposed algorithm is 90%.