HALEEM SIPINTAR MANAJEMEN

Comparative Analysis of Machine Learning Models for Heart Disease Classification: Evaluation of Random Forest, XGBoost, and Logistic Regression with Hyperparameter Optimization

2025-08-12T04:13:18+00:00

Background: Heart disease is the leading cause of death globally, necessitating an accurate early detection system using machine learning technology.

Method: This study used the UCI Heart Disease dataset with 303 samples and 14 features to compare the performance of three classification models: Random Forest, XGBoost, and Logistic Regression. Hyperparameter optimization was performed using GridSearchCV to improve model accuracy.

Results: Logistic Regression with hyperparameter optimization showed the best performance with an accuracy of 86.9%, precision of 81.3%, recall of 92.9%, and F1-score of 86.7%. The Random Forest model achieved an accuracy of 88.5%, while XGBoost achieved 85.2%.

Conclusion: Logistic Regression with hyperparameter optimization proved effective for heart disease classification with the highest recall (92.9%), which is important for early detection of heart disease, with thalach (maximum heart rate) and oldpeak (ST depression) as the main predictors.

Hybrid Deep Learning Architecture with Interpretable Feature Selection for Breast Cancer Diagnosis from Fine Needle Aspirate Images

2025-08-12T04:26:30+00:00

Background: Breast cancer is one of the leading causes of death among women worldwide. Early diagnosis plays a crucial role in improving survival rates; however, the interpretability of deep learning models often poses a significant challenge in clinical implementation.

Objective: This study aims to develop a hybrid deep learning architecture that combines Convolutional Neural Networks (CNN) with ensemble learning, equipped with feature selection techniques based on mutual information and SHAP values to improve both accuracy and interpretability in breast cancer diagnosis from Fine Needle Aspirate (FNA) images.

Method: The FNA image dataset was processed through CNN-based feature extraction, followed by feature selection using mutual information and SHAP values. An ensemble model was built using stacking that combined CNN, Random Forest, and Gradient Boosting. Evaluation was performed using accuracy, F1-score, AUC, and significance tests against the baseline model.

Results: The proposed hybrid architecture achieved an accuracy of 98.4%, an F1-score of 0.983, and an AUC of 0.995, surpassing state-of-the-art approaches that only achieved an average accuracy of 96.2%. Interpretability analysis showed that features related to texture and cell nucleus morphology had the greatest contribution to model predictions.

Conclusion: This approach not only improves the performance of FNA-based breast cancer diagnosis but also provides interpretable results for medical professionals, potentially accelerating the clinical adoption of AI-based systems.

Explainable AI-Based Heart Disease Classification: A Deep Learning Framework with Limited Features and Clinical Interpretability for Resource-Constrained Healthcare Settings

2025-08-12T04:41:00+00:00

Background: Cardiovascular disease remains the leading cause of death globally, necessitating accurate early detection systems, particularly in resource-limited areas where comprehensive diagnostic tools are unavailable. Current machine learning approaches for heart disease prediction are often clinically non-interpretative and require numerous features, limiting their application in resource-limited settings.

Objective: This study aims to develop and validate an explainable artificial intelligence (AI) framework for heart disease classification using a minimal feature set while maintaining high prediction accuracy and providing clinically understandable explanations.

Method: We implemented and compared three machine learning models (Logistic Regression, Random Forest, and Deep Neural Network) on the UCI Heart Disease dataset (n=297) with feature selection that reduced variables from 20 to 8 relevant clinical parameters. Model interpretability was enhanced using SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) techniques. Performance evaluation included AUC (Area Under the Curve), precision, recall, and F1-score metrics.

Results: Logistic regression achieved the highest performance with AUC=0.9699, precision=1.000, and recall=0.821. Random Forest showed AUC=0.9431 with balanced precision (0.917) and recall (0.786), while Deep Neural Network achieved AUC=0.9531. Clinical interpretation revealed maximum heart rate (thalach), type of chest pain (cp), and age as the most significant predictive features. The approach with limited features maintained high accuracy while reducing computational complexity by 60%.

Conclusion: This explainable AI framework demonstrates superior performance with clinical interpretability, making it suitable for implementation in healthcare facilities with limited resources. Minimal feature requirements and transparent decision-making processes enhance its practical application for early detection of heart disease and clinical decision support.

Hybrid Ensemble Learning with Active Feature Selection for Early-Stage Cardiovascular Risk Stratification: A Multi-Modal Approach Using UCI Clinical Biomarkers

2025-08-12T04:53:10+00:00

Background: Cardiovascular disease (CVD) remains the leading cause of global mortality, making accurate early-stage risk stratification crucial for optimal patient management. Traditional risk assessment methods often lack precision and fail to effectively integrate various clinical biomarkers.

Objective: This study aims to develop a hybrid ensemble learning framework with active feature selection for early-stage cardiovascular risk stratification using multi-modal clinical biomarkers.

Method: We used the UCI Heart Disease dataset (n=303) with 17 clinical features. The comprehensive methodology included active feature selection, various base models (Random Forest, XGBoost, LightGBM, Logistic Regression, SVM, KNN, Naive Bayes, Extra Trees, Gradient Boosting), and hybrid ensemble techniques (soft voting and stacked ensemble). Model evaluation was conducted using 5-fold cross-validation and SMOTE.

Results: The Logistic Regression model achieved the highest performance with an AUC of 0.9600 and an F1-score of 0.8667. The hybrid ensemble framework successfully divided patients into three risk categories: high risk (28 patients, 45.9% with an actual positive rate of 89.29%), moderate risk (10 patients, 16.4% with a positive rate of 30.00%), and low risk (23 patients, 37.7% with a positive rate of 0.00%). Cross-validation demonstrated strong performance (AUC: 0.8962 ± 0.0142, 95% CI: 0.8684–0.9240).

Conclusion: The hybrid ensemble learning approach with active feature selection provides superior accuracy in cardiovascular risk stratification, with great potential to support clinical decision-making and early intervention strategies.

Federated Learning-Based Heart Disease Prediction with Privacy-Preserving Clinical Data Fusion: A Comparative Analysis of Lightweight AI Models for Distributed Healthcare Systems

2025-08-12T05:10:58+00:00

Background: The fragmentation of health data (silos) and privacy regulations severely limit the development of robust machine learning models for heart disease prediction across distributed healthcare systems. Traditional centralized approaches require data aggregation, which raises serious privacy and regulatory concerns in clinical settings.

Objective: This study aims to develop and evaluate a privacy-preserving federated learning framework for heart disease prediction, enabling collaborative model training across healthcare institutions without compromising patient data privacy.

Method: We implemented a federated learning system using a lightweight neural network architecture across five simulated healthcare institutions (cardiology center, general hospital, emergency unit, and preventive care facility). This framework combines differential privacy mechanisms (ε = 0.718) and uses the UCI Heart Disease dataset with 303 patient records distributed among the participating institutions. Performance is compared with traditional centralized learning approaches using accuracy, precision, recall, F1-score, and AUC metrics.

Results: The federated learning model achieved an accuracy of 77.0% with an AUC of 0.6508 across all participating institutions. The performance of each institution ranged from 72.5% to 82.5% accuracy. The federated approach demonstrated a 2.0% improvement in accuracy compared to centralized learning (75.5%) while fully maintaining data locality. Communication efficiency was achieved through 8 training rounds with only 993 model parameters shared across the network.

Conclusion: Federated learning is a viable solution for privacy-preserving heart disease prediction in distributed healthcare systems, offering performance comparable to or better than centralized approaches while adhering to regulations and data sovereignty.