Artificial intelligence in myeloid malignancies: Clinical applications of machine learning in myelodysplastic syndromes and acute myeloid Leukemia.

This review summarizes applications of machine learning (ML) in acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS), spanning diagnosis, prognostication, treatment prediction, and research tools. In diagnostics, deep learning applied to bone marrow smears, peripheral blood films, and flow cytometry has shown high sensitivity and specificity, outperforming conventional methods. ML-driven unsupervised clustering and consensus classification have refined disease taxonomies, identifying genomic subtypes with prognostic value. Prognostic models and neural networks enable dynamic, personalized survival predictions. In treatment, ML assists in predicting responses to hypomethylating agents and venetoclax-based regimens, supporting clinical decision-making. In research, generative approaches create privacy-preserving synthetic cohorts and digital twins, facilitating trial design and overcoming data limitations. Future integration into clinical practice will require rigorous validation, explainable algorithms, seamless workflow incorporation, and regulatory oversight to ensure trust, equity, and safety. ML has potential to enhance multiple aspects of AML and MDS management.