Heart Disease Prediction using Logistic Regression with PCA-MFSA Feature Extraction Technique and Multidimensional Scaling (MDS) Pre-processing Approach

Heart disease prediction is critical because it has the potential to save lives, increase patient outcomes, and save the cost of healthcare. An important task in medical diagnostics is the forecasting of heart disease. In this article, we investigate the efficacy of integrating Multidimensional Scaling (MDS) preprocessing with various feature extraction methods for precise LR (Logistic Regression) heart disease prediction. The feature extraction methods under investigation include Independent Component Analysis (ICA), Principal Component Analysis (PCA) and Mean Fisher Score-based Algorithm (PCA-MFSA). PCA with MFSA emerges as the most effective method, consistently producing superior outcomes in the forms of accuracy rate, precision, and recall value compared to ICA and PCA. This underscores the significance of combining PCA with the specialized feature selection process offered by PCA-MFSA to enhance the LR model’s discriminatory power. Moreover, The incorporation of MDS preprocessing with PCA-MFSA results in a significant improvement in accuracy rate, precision value, and recall. From the result obtained PCA-MFSA produces accuracy of 91%, precision of 0.90 and recall of 0.86 respectively.