Artificial Intelligence for the Prediction of Myeloid Leukemia: A Systemic Literature Review

Mendoza-Montoya J.; Dannia Mendoza Hancco J.; Nicole Guzman Ortiz J.; Yapuchura Ocaris E. · Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology (2025) · DOI: 10.18687/laccei2025.1.1.2139

Resumen

The purpose of this Systematic Literature Review (SLR) is to analyze the impact and effectiveness of artificial intelligence (AI) models in the prediction and early diagnosis of leukemia, with a particular focus on acute myeloid leukemia (AML). Through a comprehensive review of the literature, the aim is to identify the most accurate and robust models, as well as their limitations and areas for improvement for future clinical applications. To achieve this, an SLR was conducted following the PRISMA methodology. The search was performed in the Scopus database, identifying a total of 429 articles initially. A total of 7,223 AML cases were analyzed, and various AI approaches were evaluated, including deep learning and machine learning methods. The models were assessed based on their accuracy and the quality of the data used, with emphasis on techniques such as convolutional neural networks (CNN), XGBoost, and hybrid models. The results show that the weighted convolutional neural network (WVCNN) achieved 99.9% accuracy through genomic data analysis. Techniques such as XGBoost and ResNet-50 demonstrated high effectiveness in different domains, achieving accuracy rates of 89% and 86%, respectively, depending on the type of data analyzed, whether tabular or medical images. It was concluded that AI advancements have revolutionized AML prediction by combining hybrid approaches to overcome current limitations. Therefore, the authors suggest that future research should focus on integrating multiple techniques, accessing more comprehensive and balanced datasets, and developing more interpretable models.