Abstract:
Background: AI systems and machine learning techniques hold significant potential in haematology, demonstrat-ing capabilities in diagnostic areas like cytogenetics and molecular genetics by automating blood samples and genetic data analysis Aim: This review examines the utilisation of AI systems and machine-learning techniques in haematology, focusing on their applications, potential benefits, and challenges in clinical practice. Method: This study explored various AI approaches, including Machine learning, Deep learning techniques, applied to haematological diagnostics and research through various website and journals. It evaluates the performance of AI systems in differentiating cell types, conducting chromosome banding analyses, and interpreting complex im-aging data. Results: AI systems proficiently differentiate cell types in blood samples, aiding in diagnosing condi-tions like leukemia and lymphoma. They also show promise in chromosome banding analysis, identifying ab-normalities indicative of specific haematological diseases. Machine-learning techniques have been effectively used to predict patient responses and cluster patient profiles. Discussion: Integrating AI in haematology faces challenges such as data quality, algorithm interpretability, and clinical workflow integration. The "black box" na-ture of certain AI algorithms complicates interpretability and accountability. Collaborative efforts between AI developers and clinicians are needed to address these issues. Additionally, ensuring high-quality and repre-sentative training data is crucial to avoid bias and inaccuracies in AI outcomes. Conclusion: AI shows promise in improving haematology diagnostics, but faces implementation challenges. Continued research must balance technological advancement with ethical concerns, ensuring AI supports rather than replaces haematologists.