Applications of artificial intelligence in hematology: Present and the future

TECHNOLOGICAL TRENDS IN MEDICINE

The journey began with the invention of the stethoscope in 1816; the name comes from the Ancient Greek words στῆθος (stêthos), which means “breast,” and σκοπέω (skopéō), meaning “to look.” Today, it remains a crucial instrument used by medical professionals to listen to the internal sounds of a patient’s body, including those of the heart and lungs. In the subsequent years, among many others, the ophthalmoscope (in 1851), X-ray for medical imaging (in 1895), electrocardiograph (in 1901), computed tomography (CT) scan (in 1975), positron emission tomography-CT (in 2001) were invented, and till today, they are in regular use in medicine. The first artificial intelligence (AI)-assisted medical tool, angiographic robots, was used in 2008, followed by AI-powered automatic robotic surgery in 2021.

The initial use of AI in hematology started in the 1980s in the form of image processing of blood cell morphology, followed by rule-based expert systems in laboratory medicine (1990s–2000s), Machine learning (ML) techniques, such as decision trees, were employed in 2010 in flow cytometry and genomic analysis, significantly improving the subtyping of leukemia. The concept of deep learning in the form of a convolutional neural network (CNN) designed to process data with a grid-like topology, such as images, was widely used in the late 2010–2020s. Today, we utilize generative AI to analyze molecular structures in depth for explainable insights. QProteoML is an innovative platform for quantum ML; the method takes advantage of quantum superposition and entanglement to navigate high-dimensional feature spaces more effectively than traditional algorithms. QProteoML surpasses classical ML methods (e.g., multiple myeloma) in forecasting drug resistance patterns, revealing protein signatures linked to unfavorable treatment responses. The potential benefits extend to personalized therapy choices: proteomic-based AI models could assist healthcare providers in selecting the most suitable treatment plans for individual patients.

WHY TO USE AI HEMATOLOGY?

Among the different emerging techniques in hematology, AI is going to take its place because of the following reasons:

1. Massive datasets: Hematology generates massive datasets in the areas of blood smears, flow cytometry, next-generation sequencing (NGS), multi-omics, and also in many other areas

2. Human limits: In terms of physical and psychological constraints that one health care professional face which can impact fatigue, variability and most importantly, the time constraints arising from prolonged workload, stress, demand of managing complex patient care, rushed assessment, and inadequate patient interactions may lead to decreased performance, increased risk of errors, and compromise the quality of care

3. AI excels at pattern recognition, high-dimensional processing, and predictive analytics. Due to its high ability to analyze a vast amount of data achieved through techniques such as ML, CNN, the AI systems learns from data, adapt to new inputs, and over time make powerful tools for data-driven decision making in diagnostics as well as therapeutics.

APPLICATION AREAS IN HEMATOLOGY

Hematology, which heavily depends on recognizing patterns, analyzing multiparametric data, and managing increasingly complex therapies, is an area where AI could offer significant advantages.^[1]

AI is transforming the domain of hematology with a range of applications, including:

• Diagnosis and Classification: AI technologies have been created to evaluate blood smears, bone marrow aspirates, and flow cytometry data, supporting the identification and categorization of hematologic disorders. AI is especially utilized in automated diagnostics and morphological analysis, particularly in classifying blood cells and identifying morphological irregularities

• Risk Assessment: AI tools are utilized to evaluate the likelihood of thrombosis, venous thromboembolism, and other hematologic issues

• Treatment Planning: AI aids in formulating individualized treatment strategies based on patient data and genomic insights, significantly influencing treatment planning in hematology

• Prognostication: AI algorithms are used to forecast patient outcomes and responses to treatment, enhancing clinical decision-making processes. It also serves a crucial function in flow cytometry and genomics, particularly in analyzing NGS data to detect mutations and clinically significant risk profiles

• Patient Management: AI improves the management of patients with hematologic disorders by offering insights into treatment effectiveness and patient safety

• These applications showcase how AI can enhance diagnostic precision, outcome prediction, and tailored treatment in hematology.

KEY APPLICATIONS IN 2026

• Digital Morphology: CNNs classify white blood cells with ~97% accuracy; automated bone marrow grading

• Flow Cytometry: AI detects rare clones, automates gating, achieves MRD sensitivity of 10⁻⁵

• Karyotyping: CNNs + transformers identify chromosomal abnormalities rapidly

• Liquid Biopsy: ML enhances circulating tumor DNA detection, predicts relapse 4–8 weeks earlier

• Multi-omics: GNNs integrate genomics, transcriptomics, proteomics; single-cell AI identifies stem cells with 99% accuracy

• Quantum ML: QProteoML predicts drug resistance in multiple myeloma

• AI-Prediction of Acute Leukemia: Diagnoses acute leukemias using routine laboratory parameters—no microscope needed.^[2]

AI IN HEMATOLOGY: A GADGET OR GAME CHANGER?

Despite their potential, even the most advanced tools can become ineffective if used incorrectly or if the outcomes are misunderstood. Hence, for a thorough evaluation and appropriate application of newly created AI-driven systems, it is essential for hematologists to possess a fundamental understanding of basic ML principles.^[3] The future day hematologists have to be well aware of the ML methods, their present applications and strategies across several diagnostic areas (such as cytogenetics and molecular genetics), as well as the limitations and ongoing challenges associated with these systems. Nevertheless, issues such as data quality, ethical concerns, and regulatory guidelines still need to be resolved for AI to be fully integrated into clinical practice.

A HYPE OR HOPE: DEALING WITH BOLD EXPECTATIONS AND OVERPROMISING?

AI in healthcare (including hematology care) is not a miracle – it is a collection of tools that, when used thoughtfully, can save lives, lower expenses, and enhance efficiency. This excitement often overlooks and overpowers the reality (the truth is more complex) that success relies on context, regulation, and trust. The critical factor is to differentiate between established, peer-reviewed applications (e.g., naked eye morphology under the microscope) and developing, speculative uses (like completely autonomous AI clinicians). AI is not meant to take over the role of doctors, but it is increasingly serving as a valuable tool to support them, especially in situations where speed, scalability, and pattern recognition are most crucial.

ETHICAL ISSUES AND CURRENT CHALLENGES

AI integration in hematology is currently beset by a number of ethical dilemmas. One significant obstacle to the integration of AI in hematology is the absence of transparency and explainability in numerous models, commonly known as “black boxes.”^[4] In the delicate areas of hematology treatment, especially in the management of malignancies, it is essential for healthcare providers to comprehend the rationale behind an algorithm’s predictions to uphold patient trust and guarantee medical responsibility.

THE ROADMAP

The incorporation of AI in hematology marks a major breakthrough, providing instruments for automated diagnostics, morphological evaluation, and predictive modeling. For AI to realize its full potential, it needs to fit seamlessly within clinical workflows. This necessitates the creation of user-friendly interfaces, integration with hospital information systems, and flexibility in daily operations. AI should streamline and improve medical practice rather than complicate it.