Darolia Aman, Chhillar Rajender Singh, Alhussein Musaed, Dalal Surjeet, Aurangzeb Khursheed, Lilhore Umesh Kumar
Department of Computer Science and Applications, M.D. University, Rohtak, Haryana, India.
Department of Computer Engineering, College of Computer and Information Sciences, King Saud University, Riyadh, Saudi Arabia.
Front Med (Lausanne). 2024 Jun 20;11:1414637. doi: 10.3389/fmed.2024.1414637. eCollection 2024.
Cardiovascular disease (CVD) stands as a pervasive catalyst for illness and mortality on a global scale, underscoring the imperative for sophisticated prediction methodologies within the ambit of healthcare data analysis. The vast volume of medical data available necessitates effective data mining techniques to extract valuable insights for decision-making and prediction. While machine learning algorithms are commonly employed for CVD diagnosis and prediction, the high dimensionality of datasets poses a performance challenge.
This research paper presents a novel hybrid model for predicting CVD, focusing on an optimal feature set. The proposed model encompasses four main stages namely: preprocessing, feature extraction, feature selection (FS), and classification. Initially, data preprocessing eliminates missing and duplicate values. Subsequently, feature extraction is performed to address dimensionality issues, utilizing measures such as central tendency, qualitative variation, degree of dispersion, and symmetrical uncertainty. FS is optimized using the self-improved Aquila optimization approach. Finally, a hybridized model combining long short-term memory and a quantum neural network is trained using the selected features. An algorithm is devised to optimize the LSTM model's weights. Performance evaluation of the proposed approach is conducted against existing models using specific performance measures.
Far dataset-1, accuracy-96.69%, sensitivity-96.62%, specifity-96.77%, precision-96.03%, recall-97.86%, F1-score-96.84%, MCC-96.37%, NPV-96.25%, FPR-3.2%, FNR-3.37% and for dataset-2, accuracy-95.54%, sensitivity-95.86%, specifity-94.51%, precision-96.03%, F1-score-96.94%, MCC-93.03%, NPV-94.66%, FPR-5.4%, FNR-4.1%. The findings of this study contribute to improved CVD prediction by utilizing an efficient hybrid model with an optimized feature set.
We have proven that our method accurately predicts cardiovascular disease (CVD) with unmatched precision by conducting extensive experiments and validating our methodology on a large dataset of patient demographics and clinical factors. QNN and LSTM frameworks with Aquila feature tuning increase forecast accuracy and reveal cardiovascular risk-related physiological pathways. Our research shows how advanced computational tools may alter sickness prediction and management, contributing to the emerging field of machine learning in healthcare. Our research used a revolutionary methodology and produced significant advances in cardiovascular disease prediction.
心血管疾病（CVD）是全球范围内导致疾病和死亡的普遍诱因，这凸显了在医疗数据分析领域采用精密预测方法的必要性。现有的海量医学数据需要有效的数据挖掘技术来提取有价值的见解，以用于决策和预测。虽然机器学习算法通常用于CVD的诊断和预测，但数据集的高维度给性能带来了挑战。
本文提出了一种用于预测CVD的新型混合模型，重点关注最优特征集。所提出的模型包括四个主要阶段，即预处理、特征提取、特征选择（FS）和分类。首先，数据预处理消除缺失值和重复值。随后，进行特征提取以解决维度问题，采用诸如集中趋势、定性变异、离散程度和对称不确定性等度量。使用自我改进的天鹰座优化方法对FS进行优化。最后，使用所选特征训练一个结合长短期记忆和量子神经网络的混合模型。设计了一种算法来优化LSTM模型的权重。使用特定的性能指标，将所提出方法的性能与现有模型进行评估。
对于数据集1，准确率为96.69%，灵敏度为96.62%，特异性为96.77%，精确率为96.03%，召回率为97.86%，F1分数为96.84%，马修斯相关系数为96.37%，阴性预测值为96.25%，误报率为3.2%，漏报率为3.37%；对于数据集2，准确率为95.54%，灵敏度为95.86%，特异性为94.51%，精确率为96.03%，F1分数为96.94%，马修斯相关系数为93.03%，阴性预测值为94.66%，误报率为5.4%，漏报率为4.1%。本研究的结果通过使用具有优化特征集的高效混合模型，有助于改进CVD预测。
我们通过进行广泛的实验并在包含患者人口统计学和临床因素的大型数据集上验证我们的方法，证明了我们的方法能够以无与伦比的精度准确预测心血管疾病（CVD）。具有天鹰座特征调整的QNN和LSTM框架提高了预测准确性，并揭示了与心血管风险相关的生理途径。我们的研究展示了先进的计算工具如何改变疾病预测和管理，为医疗保健领域新兴的机器学习领域做出了贡献。我们的研究采用了一种革命性的方法，并在心血管疾病预测方面取得了重大进展。
