Abstract
A brain stroke is a disruption of blood circulation to the cerebrum. As per
recent analysis, adult death and disability are primarily brought over by brain stroke.
The World Health Organization (WHO), reports that the primary cause of death and
property damage worldwide is brain stroke. Early detection of the signs and symptoms
of a stroke can help to reduce risk factor of death by up to 50%. A stroke is more likely
to occur in adults over the age of 55. An increasing number of people are experiencing
this crippling and frequently fatal form of stroke, which results in cerebral hemorrhage.
Various machine learning (ML) models were developed to predict the possibility that a
brain stroke would occur. To predict the brain stroke, the proposed system used the
CNN algorithm. The existing approaches are k-NN, Support Vector Machine (SVM),
Genetic Algorithm (GA), Naïve Bayes classifier, J48 algorithm, Logistic Regression
(LR) and Random Forest (RF). This requires more time to train the model and it is
difficult to debug. And these are not suitable for large datasets. The proposed system
makes predictions using CNN algorithm, a deep learning technique. It includes a
multilayer perceptron for the prediction task and an autoencoder for eliminating and
capturing non-linear correlations between parameters. The proposed system is
contrasted with existing system and it shows an enhancement in the capability to
anticipate the stroke. The proposed system achieved an accuracy of 89%.