Machine learning algorithms for the automated classification of pediatric anemia

Anemia is a common laboratory finding in children, caused by a multitude of common to rare and minor to life-threatening diseases. Current diagnostic tests enable early and accurate recognition of the underlying diagnosis, but their use varies with physician experience and patient setting. We investigated whether machine learning algorithms for the automated interpretation of pediatric blood counts can support the differential diagnosis of anemia. A comprehensive database of pediatric blood counts (13,913 healthy children and 1,668 pathological samples: 29.3% iron deficiency anemia, 9.2% Blackfan Diamond Anemia, 39.9% myelodysplastic syndrome, 16.4% vitamin B12-deficiency, and 5.2% heterozygous β-Thalassemia) was split into training and validation datasets to develop different machine learning models. The best performing models predicted the diagnoses in the validation dataset with an accuracy of 99.9% (classification as either “normal” or “abnormal”) and 93.9% (classification as “healthy” or specific diagnoses). This provides a proof-of-concept application of machine learning algorithms to support the diagnosis of complex hematological diseases in children.