A comparative study by using machine learning classifiers to enhance classification and prediction of heart failure disease

Abstract

PurposeHeart failure is a complex clinical condition when the heart cannot provide blood with enough flow for the body's needs. It is a major clinical and public health problem. Even if heart failure is not yet diagnosed, it is important to get your health checked every three to six months. This study aims to improve the accuracy of diagnosing heart failure by using machine learning classifiers such as Recursive Feature Elimination (RFE) and Synthetic Minority Oversampling Technique (SMOTE). MethodsHeart failure data has been acquired from the University of California, Irvine (UCI) repository. To improve the accuracy of diagnosing heart failure, we employed the following methods for this study: k-Nearest Neighbor, Naive Bayes (NB), Random Forest, XGBoost, Decision Tree (DT), Logistic Regression (LR), and Support Vector Machines (SVM). The model was validated using the F-measure and ROC-AUC (Receiver Characteristic Area Under Curve) methods. ResultsSupport vector machines employing logistic regression as a feature selection strategy produced the most significant classification accuracy of 90%, while support vector machines utilising RF as a feature selection strategy showed an accuracy of 83%. We also have an accuracy of 90% in the random forest as a machine learning methodology with all of our features.ConclusionThe small dataset size of the current research presents a challenge to everyone's ability to gain more accurate findings. Improved diagnostics for heart failure may be possible in the future using our machine-learning classifier-based classification system. To accurately forecast heart failure, this is the easiest way to use and the most accurate.

Heart Failure Prediction Using Machine Learning

“…Traditional biostatistics analyses have also affirmed the importance of feature ranking carried out via machine learning [9]. Employing techniques such as Recursive Feature Elimination (RFE) with Random Forest and Logistic Regression has led to improved model performance [10]. Notably, studies have found that utilizing a subset of features yields better performance compared to using all available features [10].…”

“…Employing techniques such as Recursive Feature Elimination (RFE) with Random Forest and Logistic Regression has led to improved model performance [10]. Notably, studies have found that utilizing a subset of features yields better performance compared to using all available features [10]. The ROC curve visually illustrates the trade-off to between the true positive rate and the false positive rate for a binary classification model across different classification thresholds, aiding in model evaluation and comparison [11].…”