Scholarlink

<p>In the field of healthcare analytics, reliable illness prediction and decision-making are dependent on comprehending the complex relationships between symptoms and inter-attribute linkages. The main objective of the proposed methodology uncovers these inter-attribute relationships by combining association rule generation with hybrid optimization algorithms. This work goes beyond cardiovascular problems to address a variety of health challenges by employing benchmark datasets such as the UCI repository’s Heart Disease Prediction dataset. These datasets, together with others that cover a variety of health issues, serve as the foundation for developing rules. To improve the effectiveness and precision of association rule generation, the proposed approach hybridize the Butterfly Optimization Algorithm (BOA) with Moth-Flame (MFOA), Coati Optimization (COA) and Pelican (POA). By combining the traditional FP-growth and Apriori algorithms, the suggested optimization methodology efficiently discovers common itemsets and reveals complicated relationships between various health variables. The antecedent and consequent symptoms of each disease are evaluated by using fitness score, support, confidence and lift factors. The rules of BOA is passing to all hybridized algorithms and BOA + POA produces the best fitness score, support and confidence compared to other algorithms. The processing time of each algorithm is evaluated for heart disease, diabetes and parkinsons datasets and BOA + POA takes less time for various population size and iterations. The POA significantly improves the performance, increasing scalability, fitness by 11% and confidence by 40% for various diseases.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Effective Association Rule Generation in Healthcare Using Optimization Algorithms

  • 期刊论文

  • 发表日期：

    2026年3月30日

查看全文
  • Santhi Selvaraj,
  • S. Selva Nidhyananthan,
  • T. Rama Subramanian,
  • T. R. Trilok Guru

摘要

In the field of healthcare analytics, reliable illness prediction and decision-making are dependent on comprehending the complex relationships between symptoms and inter-attribute linkages. The main objective of the proposed methodology uncovers these inter-attribute relationships by combining association rule generation with hybrid optimization algorithms. This work goes beyond cardiovascular problems to address a variety of health challenges by employing benchmark datasets such as the UCI repository’s Heart Disease Prediction dataset. These datasets, together with others that cover a variety of health issues, serve as the foundation for developing rules. To improve the effectiveness and precision of association rule generation, the proposed approach hybridize the Butterfly Optimization Algorithm (BOA) with Moth-Flame (MFOA), Coati Optimization (COA) and Pelican (POA). By combining the traditional FP-growth and Apriori algorithms, the suggested optimization methodology efficiently discovers common itemsets and reveals complicated relationships between various health variables. The antecedent and consequent symptoms of each disease are evaluated by using fitness score, support, confidence and lift factors. The rules of BOA is passing to all hybridized algorithms and BOA + POA produces the best fitness score, support and confidence compared to other algorithms. The processing time of each algorithm is evaluated for heart disease, diabetes and parkinsons datasets and BOA + POA takes less time for various population size and iterations. The POA significantly improves the performance, increasing scalability, fitness by 11% and confidence by 40% for various diseases.