Brain Stroke Prediction Using Logistic Regression with Logarithmic Transform

Stroke poses a significant burden on individuals and healthcare systems globally, highlighting the crucial need for timely identification and prediction of stroke risk factors. Predictive modelling through data science offers a promising approach for enhancing our understanding of stroke risk factors and improving the accuracy of stroke prediction. This research focuses on binary classification outcomes, with logistic regression emerging as a potential solution to address the challenges of stroke prediction in healthcare. By emphasizing data-driven methods to identify individuals at higher risk, stroke prediction has the capacity to reform healthcare by facilitating early intervention and personalized preventive measures. Logarithmic transformations of non-linear features are performed in model to improve binary prediction accuracy. The modified model performs better with an AUC value of 0.92 as compared to the basic model with an AUC value 0.91.