Scholarlink

<p>Anemia remains a critical public health burden across South Asian countries such as Bangladesh and India, affecting women of reproductive age, young children, and teenagers. Its widespread prevalence highlights the need for efficient and accessible screening approaches that can support early detection and timely intervention. In this study, we introduce an XAI-driven framework designed to detect anemia using routine hematological parameters. Leveraging clinical records from Bangladeshi patients, we developed an ensemble model that integrates five machine learning algorithms (Decision Tree Classifier, Logistic Regression, Support Vector Classifier, K-Nearest Neighbors, and Gaussian Naïve Bayes), enhanced with interpretability tools to support clinical decision-making. To evaluate the robustness of our pipeline, we conducted cross-dataset validation using an independent clinical dataset from India as well as an open-access dataset from Kaggle. Across both datasets, the proposed framework demonstrated consistently strong performance, reinforcing its adaptability and generalizability. We further performed risk factor analysis and risk stratification across different age groups and genders. Statistical evaluations, supported by SHAP-based explanations, identified ‘Hemoglobin,’ ‘Packed Cell Volume,’ and ‘Red Blood Cell count’ as the most influential risk factors for anemia in the Bangladeshi population. Besides, anemia risk showed a gradual increase with age (Odds Ratio = 1.02, 95% CI: 1.01–1.03, <i>p</i> &lt; 0.001) and remained notably higher among female subjects (Odds Ratio = 0.17, 95% CI: 0.13–0.23, <i>p</i> &lt; 0.001). To support practical clinical use, we designed a user-friendly real-time prediction interface that includes clear model explanations and risk probability. This system can serve as an interpretable decision support tool that helps healthcare providers understand predictions while enabling automated anemia assessment. Thus, it can improve screening coverage, minimize diagnostic delays, and ultimately enhance anemia management across Bangladesh and similar population settings.</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

Risk Factors Analysis and Early Detection of Anemia from Clinical Hematological Parameters

  • 期刊论文

  • 发表日期：

    2026年4月4日

查看全文
  • Pankaj Bhowmik,
  • Mostofa Kamal Nasir

摘要

Anemia remains a critical public health burden across South Asian countries such as Bangladesh and India, affecting women of reproductive age, young children, and teenagers. Its widespread prevalence highlights the need for efficient and accessible screening approaches that can support early detection and timely intervention. In this study, we introduce an XAI-driven framework designed to detect anemia using routine hematological parameters. Leveraging clinical records from Bangladeshi patients, we developed an ensemble model that integrates five machine learning algorithms (Decision Tree Classifier, Logistic Regression, Support Vector Classifier, K-Nearest Neighbors, and Gaussian Naïve Bayes), enhanced with interpretability tools to support clinical decision-making. To evaluate the robustness of our pipeline, we conducted cross-dataset validation using an independent clinical dataset from India as well as an open-access dataset from Kaggle. Across both datasets, the proposed framework demonstrated consistently strong performance, reinforcing its adaptability and generalizability. We further performed risk factor analysis and risk stratification across different age groups and genders. Statistical evaluations, supported by SHAP-based explanations, identified ‘Hemoglobin,’ ‘Packed Cell Volume,’ and ‘Red Blood Cell count’ as the most influential risk factors for anemia in the Bangladeshi population. Besides, anemia risk showed a gradual increase with age (Odds Ratio = 1.02, 95% CI: 1.01–1.03, p < 0.001) and remained notably higher among female subjects (Odds Ratio = 0.17, 95% CI: 0.13–0.23, p < 0.001). To support practical clinical use, we designed a user-friendly real-time prediction interface that includes clear model explanations and risk probability. This system can serve as an interpretable decision support tool that helps healthcare providers understand predictions while enabling automated anemia assessment. Thus, it can improve screening coverage, minimize diagnostic delays, and ultimately enhance anemia management across Bangladesh and similar population settings.