ARTIFICIAL INTELLIGENCE IN HEMATOLOGY: APPLICATIONS IN THE DIAGNOSIS, PROGNOSIS, AND TREATMENT OF HEMATOLOGICAL MALIGNANCIES

Abstract

Hematological malignancies are biologically heterogeneous disorders in which accurate classification and timely treatment depend on the integration of morphology, immunophenotyping, cytogenetics, molecular profiling, and clinical assessment. Artificial intelligence, including machine learning and deep learning, offers methods for extracting clinically relevant patte rns from these increasingly complex data. This narrative review summarizes recent advances in AI -assisted diagnosis, prognostic assessment, and treatment support in leukemia, lymphoma, and multiple myeloma. Digital image analysis can support cell recognition and tissue classification; algorithms applied to genomic and clinical data may refine risk estimation; and multimodal models may help characterize treatment response and toxicity. Nevertheless, performance reported in retrospective datasets does not automatically translate into clinical benefit. Differences in instruments, staining, patient populations, data quality, and clinical workflows can impair generalizability, while limited interpretability and automation bias create safety concerns. Current evidence supports AI as an adjunct to, rather than a replacement for, hematopathologists and hematologists. Prospective external validation, transparent reporting, privacy-preserving data governance, and continuous monitoring are prerequisites for responsible implementation.