What is the clinical evidence from this study?

Study design: Other. Population: Heart failure (n=299). Intervention: Random Forest (RF) algorithm vs. KNN, DT, NB, SVM, and LR algorithms. Primary outcome: Accuracy of heart failure prediction.

What does this research mean for the field?

A machine learning-based decision support system utilizing the Random Forest algorithm with standardization achieves 90% accuracy in predicting heart failure, outperforming other tested models. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

The aim is to develop a machine learning-based model for accurate detection of heart failure.

May 4, 20260 citationsOpen Access

A Machine Learning Based Decision Support System for Predictive Detection of Heart Failure

SBSema Bilekyiğit AEAyşe Eldem

Key Result

The Random Forest algorithm with standardization and 100 trees achieved the highest accuracy of 90% in predicting heart failure compared to other machine learning models.

Key Points

The aim is to develop a machine learning-based model for accurate detection of heart failure.
Utilized heart failure dataset from UCI Machine Learning Repository
Applied preprocessing steps including normalization, standardization, and Recursive Feature Elimination
Implemented machine learning algorithms like RF, KNN, SVM, DT, NB, and LR with 5-fold cross validation
RF algorithm achieved 90% accuracy in predicting heart failure
Other algorithms demonstrated varying performance, but RF was the most effective
The proposed model aids in enhancing diagnostic accuracy for heart failure using explainable AI techniques

Structured PICO

Does a machine learning-based decision support system accurately predict death or survival in heart failure patients?

Population

299 sample patient data from the UCI Machine Learning Repository, age range 50-90

Intervention

Machine learning algorithms (Random Forest, K-Nearest Neighbor, Decision Trees, Support Vector Machine, Naive Bayes, Logistic Regression) with preprocessing steps (normalization, standardization, Recursive Feature Elimination, Logistic Regression)

Comparator

Comparison between different machine learning algorithms and preprocessing techniques (unscaled vs. normalized vs. standardized)

Outcome

Accuracy in predicting death or survival status

A Random Forest machine learning algorithm with data standardization achieved 90% accuracy in predicting mortality in heart failure patients, offering a potential decision support tool for clinicians.

Limitations

Black box nature of machine learning analyses

Abstract

Heart failure is one of the most chronic diseases in recent years. In this disease, structural and functional disorders are seen in filling or pump functions of the heart. In order to eliminate the error factor of people, especially in the diagnosis of disease, models have been developed that make predictions close to the truth with Machine Learning Algorithms (MLAs). In this study, a new model with different pre-processing steps was proposed to provide the highest accuracy in the detection of heart failure disease. The preprocessing steps such as normalization, standardization, Recursive Feature Elimination (RFE) and Logistic Regression (LR) were applied to the heart failure dataset from UCI Machine Learning Repository. After that, MLAs such as K-Nearest Neighbor (KNN), Decision Trees (DT), Support Vector Machine (SVM), Random Forest (RF), Naive Bayes (NB) and LR were applied with 5-fold cross validation. According to the results of the experiments, the best performance was obtained from the RF algorithm with 90% accuracy. The predictions of the proposed machine learning-based decision support system, which is interpreted with explainable artificial intelligence, will help doctors diagnose heart patients more effectively.

问 AI

Bookmark

View Full Paper