A Thorough Examination of a Stroke Detection using Deep Learning Models

Brain stroke or paralysis is a devastating condition for human health, leaving the patient completely numb for the remainder of their life and causing complete trauma to the patient’s family members while they are still alive. Early brain stroke prediction is a challenging task for the medical field, as it requires continuous evaluation of the disease through numerous tests. Artificial intelligence plays a vital role in disease prediction and detection through its high-end models, like deep learning. Most deep learning models work on a single type of data, such as brain stroke neural images or statistical data related to the disease. This single neural network is insufficient for accurately predicting and detecting brain stroke possibilities. Therefore, the proposed model leverages heterogeneous data, such as images and statistical data, to accurately predict and detect the risk of brain stroke. To achieve this, the proposed model uses neural images to train a tailor-made channel boost convolution neural network, along with a non-negative matrix factorization technique, which catalyses the identification of the stages of brain stroke proneness. Conversely, we employ statistical data to identify potential brain strokes through a long-short-term neural network methodology. The outcomes from both neural networks are combined to enhance results using a decision tree model.