BACKGROUND
Childhood leukemia is one of the most prevalent forms of pediatric cancer and occurs in two forms; Acute Lymphoblastic Leukemia (ALL) and Acute Myeloid Leukemia (AML). Prompt treatment of these ailments has been shown to significantly improve the survival rate of children afflicted with acute leukemia.
OBJECTIVE
In an effort to develop an early and comprehensive predictor of hematologic malignancy in children.
METHODS
we studied nutritional markers, key indicators of leukemia, and granulocytes in the patient's blood. Applying a machine learning algorithm and ten indices, our team analyzed 826 pediatric patients with ALL and 255 children with AML, comparing them with a control group of 200 healthy children.
RESULTS
The study uncovered noteworthy distinctions between boys and girls, as well as the relationship between blood biochemical markers. Through the use of a random forest model, we achieved an Area Under the Curve (AUC) of 0.950 for the prediction of leukemia subtypes, and an AUC of 0.909 for the prediction of AML.
CONCLUSIONS
This research provides an efficient and auxiliary diagnostic tool for the early screening of childhood blood cancers, and demonstrates the potential of artificial intelligence in modern healthcare.