Revolutionizing anemia detection: integrative machine learning models and advanced attention mechanisms

Area of Science:

• Biomedical Engineering
• Computational Biology
• Medical Informatics

Background:

• Anemia is a prevalent blood disorder with significant health consequences, often undetected due to limitations in traditional diagnostic methods.
• Current diagnostic approaches for anemia are time-consuming and subjective, highlighting the need for efficient and objective detection techniques.
• Machine learning (ML) offers a promising avenue for developing automated and accurate anemia detection systems.

Purpose of the Study:

• To explore the efficacy of various machine learning classification models for anemia detection.
• To investigate the performance of innovative ML models incorporating attention mechanisms and spatial attention for improved anemia diagnosis.
• To evaluate an integrated approach combining textual and image data for noninvasive anemia detection.

Main Methods:

• Application of diverse ML classification algorithms including logistic regression, decision trees, random forest, support vector machines, Naïve Bayes, and k-nearest neighbors.
• Development and evaluation of advanced models featuring attention modules and spatial attention for enhanced feature extraction.
• Utilized both textual and image datasets, along with an integrated multimodal approach, for anemia detection.
• Conducted ablation studies to ascertain the contribution of individual model components, such as blue-green-red, multiple, and spatial attentions.

Main Results:

• Proposed ML models demonstrated high accuracy, precision, recall, and F1 scores across textual and image datasets.
• The integrated approach combining textual and image data significantly outperformed single-modality analyses.
• The AlexNet Multiple Spatial Attention model achieved a remarkable accuracy of 99.58% in anemia detection.
• Ablation studies confirmed the critical role of attention mechanisms in boosting model performance.

Conclusions:

• The study presents a comprehensive and innovative framework for noninvasive anemia detection using advanced ML techniques.
• The developed models, particularly the AlexNet Multiple Spatial Attention model, show significant potential for revolutionizing automated anemia diagnosis.
• The findings underscore the effectiveness of multimodal data integration and attention mechanisms in improving the accuracy and efficiency of anemia detection systems.