<![CDATA[Anemia is a global health problem affecting approximately 1.92 billion people, or 24% of the population, according to the WHO. Accurate early detection is crucial for data-driven healthcare. This study evaluates two machine learning algorithms, Gaussian Naive Bayes (GNB) and Random Forest (RF). The classification is based on four key hematological parameters: Hemoglobin (Hb), Mean Corpuscular Volume (MCV), Mean Corpuscular Hemoglobin (MCH), and Mean Corpuscular Hemoglobin Concentration (MCHC). GNB relies on Bayes' Theorem with Gaussian distribution assumptions, whereas RF is a decision-tree-based ensemble method capable of capturing non-linear patterns without specific distributional assumptions. Evaluated using 5-fold cross-validation and standard metrics (accuracy, precision, recall, F1-score), results showed that RF outperformed GNB. RF achieved 94.2% accuracy (CV 94.9% ± 1.1%), compared to GNB's 90.5% (CV 90.1% ± 1.3%). RF feature importance confirmed Hb as the dominant predictor (score 0.562), aligning with its strong correlation to anemia (r = −0.80). Although not surpassing larger-scale studies, these results remain highly competitive. Ultimately, this research provides evidence to support the development of automated, data-driven clinical decision support systems for anemia detection.]]>
Copyrights © 2026
