A novel dynamic Optuna hybrid Harris Hawks Optimization approach for classification of CAD

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: InceptionV3 with Dynamic Optuna H-HHO achieved 97.9% accuracy; MobileNet achieved 97.6% accuracy. Conventional H-HHO maximum accuracy was 82.46% — an absolute improvement of approximately 15.4 percentage points for the best-performing model

• Primary Outcome: Classification accuracy for CAD prediction using deep learning models optimised with Dynamic Optuna H-HHO versus conventional H-HHO

• Nnt Or Sensitivity: Sensitivity, specificity, AUC-ROC, PPV, and NPV not reported in abstract. Diagnostic accuracy metrics required for clinical utility assessment are absent

• Confidence Interval: Not reported

Clinical Application

Deep learning-based CAD classification tools are technically feasible in principle and align with emerging digital health infrastructure. However, this specific framework lacks the clinical validation, regulatory approval pathway, and implementation evidence required for clinical deployment. Computational requirements for running ResNet-50, VGG-16, InceptionV3, or MobileNet with the proposed optimisation framework in a clinical setting are not discussed. In Australia, CAD diagnosis is guided by NHFA/CSANZ guidelines and typically involves clinical risk stratification, exercise stress testing, CT coronary angiography (CCTA), and invasive coronary angiography where indicated. AI-assisted diagnostic tools for cardiac imaging are an active area of development, but any such tool would require TGA regulatory approval as a Software as a Medical Device (SaMD) under the TGA's Digital Health framework. The tool would need to demonstrate clinical validity in Australian patient populations. There is no PBS pathway for AI diagnostic software at present. RACGP and CSANZ would require prospective clinical validation data before endorsing such tools in primary or specialist care. This study, in its current form, does not meet the evidentiary threshold for any of these pathways. The study targets patients with suspected CAD requiring classification of coronary artery stenosis. However, the uncharacterised dataset and absence of external validation mean the model cannot currently be recommended for any specific clinical population.

Abstract

Coronary Artery Disease (CAD) is a leading cause of mortality worldwide and is primarily associated with atherosclerotic plaque formation, resulting in coronary artery stenosis. The accurate prediction of CAD using deep learning models is often constrained by the limitations of conventional optimization techniques, including premature convergence and limited adaptability of the models. To address these challenges, this study proposes a Dynamic Optuna Hybrid Harris Hawks Optimization (Dynamic Optuna H-HHO) framework to enhance the performance of deep learning-based CAD prediction models. The proposed approach integrates dynamic parameter adjustment, adaptive escape energy mechanisms, and Optuna-based hyperparameter tuning. The framework was applied to optimize several deep-learning classifiers, including ResNet-50, VGG-16, InceptionV3, and MobileNet, using a coronary artery stenosis dataset. The performance was evaluated through a comparative analysis with models optimized using the conventional Hybrid Harris Hawks Optimization (H-HHO) algorithm. The experimental results indicate that the proposed Dynamic Optuna H-HHO framework consistently improves the predictive accuracy across all evaluated models. InceptionV3 achieved the highest accuracy of 97.9%, followed by MobileNet with 97.6%, compared with the maximum accuracy of 82.46% obtained using traditional HHO-based optimization. By combining adaptive optimization strategies with automated hyperparameter tuning, the proposed framework provides a robust and scalable solution for improving the accuracy of coronary artery disease prediction.

References

1. 1.Revathi, C. K., & Santhi, H. (2026). A novel dynamic Optuna hybrid Harris Hawks Optimization approach for classification of CAD. PLoS ONE. https://doi.org/10.1371/journal.pone.0353605