Biased Cv Static Correlation Based XGB-SVM Stack Model for Stroke Prediction

Abstract

Abstract The earlier a stroke is detected, the better the odds of successful treatment and recovery. Early identification of people at high risk of stroke can lead to the implementation of preventative interventions, lowering the incidence of stroke. Machine learning has emerged as a valuable technique in stroke prediction. It can be used to create risk prediction models, analyse medical pictures, real-time monitor patients, create personalised treatment plans, and anticipate the likely result of a stroke. Machine learning can assist healthcare providers in identifying patients at high risk of stroke, monitoring them in real time, and developing personalised treatment regimens to reduce stroke incidence and improve outcomes. Nave Bayes (NB), Logistic Regression, and other machine learning algorithms are used to predict strokes. Machine learning algorithms such as Nave Bayes (NB), Logistic Regression (LR), Support Vector Machine (SVM), Decision Tree (DT), Extreme Gradient Boosting (XGB), K closest Neighbours (KNN), and Random Forest (RF) are used to predict strokes.According to the observations, the model performs poorly in terms of accuracy and other metrics.The biassed Cramer's V (CV) static correlation with XGB-SVM stack model is proposed to solve the problem. To avoid overfitting, data pre-processing is used first to eliminate null values and oversample the given dataset. Biassed CV static correlation is used in the initial step to minimise the dimensionality of the dataset and choose features based on the important factor of each variable. The XGB-SVM stack model is utilised to predict the stroke efficiently in the second phase. According to the observations, the accuracy of the model is 97.6%, which is exceptionally high when compared to other models. When the Area under Curve (AUC)-Receiver Opportunistic Curve (ROC) is 0.99, it shows that the classification's performance is faultless.