Enhancing Accuracy and Efficiency in Heart Disease Prediction for Healthcare

Cardiovascular diseases still happen to be one of the highest causes of mortality globally, developing precise and interpretable prediction models has become essential more than ever. This research paper introduces advanced machine learning techniques to forecast heart disease with the help of heart_disease_data.csv dataset. The proposed method overcomes the shortcomings of previous studies, including inadequate feature selection, class imbalance and limited model interpretability. The proposed method for the development of the model is to pre-process the dataset. For optimal accuracy the dataset was scaled with StandardScaler, the class imbalance was handled through SMOTE. Recursive feature elimination was utilized to identify the ten most prominent features. A hybrid model was created by integrating logistic regression, gradient boosting and random forest within a soft voting classifier framework. The developed model achieved a precision rate of 84.85% that surpasses many recent studies that reported accuracy ranging from 75% to 83.6%. By evaluating other performance metrics, like recall (90.91%), precision (81.08%), F1-score (85.71%) and ROC AUC (0.85), further confirm the model’s reliability. In summary, the combination of feature selection, class balancing, ensemble learning, significantly improves the accuracy and relevance of predictions in clinical settings.