In a recent study published in Scientific Reports, researchers developed a machine learning-based heart disease prediction model (ML-HDPM) that uses different combinations of information and a number of recognized classification methods.

study: Comprehensive evaluation and performance analysis of machine learning in heart disease prediction. Image credit: Summit Art Creations/Shutterstock.com

background

Heart disease is a global health risk that medical professionals must assess and treat through physical exams, advanced imaging techniques, and diagnostic procedures. Promoting heart-healthy habits and early diagnosis can help minimize the incidence of cardiovascular disease and improve overall health.

Current approaches such as machine learning, deep learning, and sensor-based data collection yield promising results but have limitations such as uneven diagnostic accuracy and overfitting.

The proposed approach uses modern technology and feature selection procedures to enhance the diagnosis and prognosis of heart disease.

About research

In the current study, researchers built an ML-HDPM model for accurate heart disease prediction.

The researchers used the Cleveland database, Swiss database, Long Beach database, and Hungarian database to obtain cardiovascular data. They preprocessed clinical data, followed by feature selection, feature extraction, cluster-based oversampling, and classification.

They used the training data to fit the model to the feature set, computed the importance score, and removed the lowest feature score to achieve the desired feature.

The genetic algorithm (GA) included population initialization, selection, crossover, and mutation to determine whether termination criteria were met.

Researchers undersampled raw data samples with majority labels and clustered samples with minority labels, merged training sets, and performed synthetic minority oversampling (SMOTE) to generate model output. did.

The model uses Recursive Feature Elimination Method (RFEM) and Genetic Algorithm (GA) to select relevant features and improve model resilience. Techniques such as the Undersampling Clustering Oversampling Technique (USCOM) correct for data imbalance.

The classification task uses a multilayer deep convolutional neural network (MLDCNN) and adaptive elephant swarm optimization method (AEHOM).

The model classifiers were Principal Component Analysis (PCA), Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), Decision Tree (DT), Random Forest (RF), and Naive Bayes (NB).

This model combines supervised infinite feature selection with an upgraded weighted random forest algorithm. The preprocessing steps of ML-HDPM ensure data integrity and model validity. Extensive feature selection reveals important properties for predictive modeling.

Scalar methods achieve consistent feature effects, and SMOTE compensates for class imbalance. Genetic algorithms employ the principle of natural selection to generate multiple solutions in one generation.

The performance of the strategy is evaluated through simulation tests and compared with existing models. The test, training, and validation datasets consisted of 80%, 10%, and 10% data, respectively.

result

As evidenced by comprehensive testing, ML-HDPM performed admirably across a wide range of important evaluation criteria. The ML-HDPM model predicted cardiovascular disease with 96% accuracy and 95% accuracy using training data.

The sensitivity (recall) of the system yields an accuracy of 96%, and the F-score of 92% reflects a balanced performance. The ML-HDPM specificity of 90% is noteworthy.

ML-HDPM provides accurate and reliable results. It incorporates complex technologies such as feature selection, data balancing, deep learning, and Adaptive Elephant Husbandry Optimization (AEHOM). These strategies enable models to reliably predict heart disease, improving clinical decision-making and patient outcomes.

ML-HDPM outperforms other algorithms in training (95%) and testing (88%). This success is due to a combination of complex feature extraction, data imbalance correction, and machine learning.

Feature selection algorithms can find important qualities related to cardiovascular health and help detect subtle patterns indicative of cardiovascular disease.

Data correction using efficient data balancing techniques ensures model training on representative datasets, such as deep learning using the MLDCNN approach and AEHOM optimization to improve model accuracy.

ML-HDPM, a deep learning model, has a lower false positive rate (FPR) in training (8.20%) and testing (15%) than other approaches due to improvements in feature selection, data balance, and machine learning components of ML-HDPM. ) will be lower. .

The model showed high true positive rate (TPR) on training (96%) and test (91%) datasets due to improvements in feature identification, data balance, and deep learning. This approach improves the model’s ability to identify true positives.

conclusion

This study introduces a unique ML-HDPM approach that incorporates feature selection, data balance, and machine learning to improve prediction of cardiovascular disease.

The balanced F-measure of precision and recall, high precision and precision, and low false positive rate on training and test datasets highlight the promising potential of the model in cardiovascular diagnostic applications.

The findings indicate that the ML-HDPM model can improve the accuracy and speed of cardiovascular disease identification, potentially improving standard of care.

However, further research is needed to improve model optimization and data quality and explore model use by healthcare professionals in real-world settings.