An AI-Based Decision Support Framework for Clinical Screening of Iron Deficiency Anemia and Thalassemia

Iron deficiency anemia (IDA) and thalassemia (THL) are common hematological disorders that necessitate efficient and accurate screening for early diagnosis. Traditional blood smear analysis is labor-intensive, prone to subjectivity, and lacks reproducibility, highlighting the urgent need for AI-driven methods to improve diagnostic sensitivity and specificity. This study proposes a novel hybrid AI framework for patient-level classification, integrating soft voting with optimized probability-based thresholds. The model was trained and validated using a real-world dataset from Hatyai Hospital, Thailand, and evaluated at both the patch and patient levels. The proposed approach achieved 96% accuracy on the test set. Specifically, it yielded precision-recall values of 1.00 and 0.83 for IDA, and 0.95 and 1.00 for THL, respectively. At the patient level, sensitivity reached 1.00 for THL and 0.83 for IDA. Bayesian probability updates further confirmed prediction reliability, yielding post-test probabilities exceeding 99.99% for IDA and 80% for THL. The model explained 62.84% of the variance in patient classifications, demonstrating strong discriminatory power. Model interpretability, assessed using SHAP and Grad-CAM, highlighted key red blood cell morphological features. The proposed framework thus serves as a cost-effective screening tool. Limitations include the use of a single-center dataset and the need for adaptive threshold optimization. Future work will focus on multi-center validation and real-world clinical integration. This study thereby establishes a structured baseline for AI-assisted hematology screening, fostering early detection and improved clinical decision-making in resource-limited settings.