# Digital Medicine and Artificial Intelligence in Chronic Myeloid Leukemia: Current Applications, Challenges, and Future Directions

## TL;DR

AI and digital tools are transforming CML care by improving diagnosis, treatment, and personalized medicine, but challenges like data quality and patient trust must be addressed.

## Contribution

This review systematically examines AI applications in CML and highlights the need for ethical, validated, and patient-centered AI integration.

## Key findings

- AI/ML improves diagnostic accuracy in hematology through cytomorphology and flow cytometry.
- Multimodal ML frameworks enhance disease classification and outcome prediction in CML.
- Robust validation and MLDevOps infrastructure are critical for safe AI clinical integration.

## Abstract

Chronic myeloid leukemia (CML) has become a paradigm for targeted therapy with BCR-ABL1 tyrosine kinase inhibitors (TKIs). However, the growing volume and complexity of clinical, molecular, and imaging data challenge traditional decision-making based on static risk scores. Digital health technologies and artificial intelligence (AI) offer new opportunities to enhance diagnosis, risk stratification, and treatment personalization in CML.

This narrative review is based on a focused literature search in PubMed/MEDLINE and Web of Science (2010-2025), combined with expert selection of pivotal studies in CML, digital health, and AI. We included peer-reviewed original research and reviews describing applications of digitalization, AI, or machine learning (ML) in CML or closely related hematologic malignancies, as well as key publications on ethics, regulation, patient perspectives, and ML operations (MLDevOps).

We summarize the integration of electronic health records, telemedicine, networked registries, and real-world evidence as a foundation for AI in CML. We review AI/ML applications in diagnostic hematology (cytomorphology, flow cytometry, cytogenetics, histopathology), prognostic modeling, molecular response monitoring (including automated BCR-ABL1 trend analysis and ghost cytometry), drug discovery, and clinical decision support systems (CDSS). Multimodal ML frameworks that integrate clinical, imaging, histopathological, and genomic data enable more precise disease classification and outcome prediction. At the same time, we discuss challenges related to data quality, algorithmic bias, model transparency, regulatory oversight, and patient trust, emphasizing the need for robust validation and MLDevOps infrastructure.

AI has the potential to substantially improve CML diagnosis, prognostication, and treatment selection and to support innovative approaches such as treatment-free remission and rational drug design. However, technical sophistication alone is insufficient. Safe and effective clinical integration of AI in CML will require rigorous multicenter validation, continuous performance monitoring, explainable models aligned with ELN guidelines, appropriate regulatory frameworks, and patient-centered implementation strategies. Under these conditions, AI can become a key enabler of truly personalized, evidence-based, and patient-centered care in CML.