Diagnosis and classification of thalassemia disease using machine learning: Comparative analysis of traditional models and a novel hybrid approach
Machine learning model diagnoses thalassemia subtypes from standard blood tests with 83% accuracy, potentially reducing diagnostic delays and treatment errors.
A novel hybrid machine learning model (ThalP) combining SVM, logistic regression, and XGBoost achieves 83.1% accuracy for thalassemia subtype classification using only routine complete blood count parameters — a clinically important capability given that diagnostic errors between alpha/beta thalassemia subtypes affect treatment planning. This is the first result in T2 (CBC-based ML in hematology) after 4 consecutive zero-result runs for this topic.
What the study was
- Study design
- ML model development and external validation on clinical dataset
- Population
- Thalassemia patients evaluated at Atatürk University Hematology Department
- Sample size
- 349
- Category
- Diagnostics
- Maturity
- Validated
- Journal
- Technology and Health Care
Why it surfaced
Addresses T2 watchlist topic (CBC+ML hematology) which had zero results for 4 prior consecutive runs; hybrid stacking model with external validation is methodologically sound; thalassemia is a globally significant hereditary blood disorder with diagnostic challenges in resource-limited settings.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.