Score and Correlation Coefficient-Based Feature Selection for Predicting Heart Failure Diagnosis by Using Machine...

Area of Science:

• Cardiology
• Machine Learning
• Data Science

Background:

• Cardiovascular disease (CVD) is a leading cause of mortality, with heart failure (HF) being a primary concern.
• Current HF diagnosis relies on electronic medical records but is challenged by resource and expertise limitations.
• Accurate and timely HF prediction is crucial for patient outcomes and healthcare efficiency.

Purpose of the Study:

• To develop a machine learning approach for predicting heart failure (HF) using electronic clinical records.
• To identify and engineer the most relevant features from patient data for improved diagnostic accuracy.
• To evaluate the performance of various classification algorithms in predicting HF.

Main Methods:

• Applied SelectKBest with chi-squared statistics to identify key features in electronic health records.
• Utilized feature engineering to create new, strongly correlated features for model training.
• Trained and evaluated optimized hyperparameter classification models including SVM, KNN, Decision Tree, Random Forest, and Logistic Regression on two datasets.

Main Results:

• The Random Forest algorithm demonstrated superior performance, achieving 95% accuracy, 97.62% precision, 95.35% recall, and 96.47% F1 score on the HF prediction dataset.
• Random Forest achieved 100% accuracy on the Cleveland dataset and 97.68% on the second dataset.
• The Random Forest model attained 100% precision, recall, and F1 scores for both datasets, indicating robust predictive power.

Conclusions:

• Machine learning, particularly the Random Forest algorithm, offers a highly effective method for predicting heart failure (HF) from electronic clinical records.
• The proposed feature selection and engineering approach enhances the predictive accuracy of machine learning models for cardiovascular disease.
• This data-driven strategy can significantly aid in early diagnosis and management of heart failure, overcoming traditional diagnostic challenges.