Authors - Sarita Thummar, Amit Thakkar, Gayatri Patel, Vaishali Koria, Yug Mordiya Abstract - Leukemia is a malignancy that afflicts blood and bone marrow and requires a precise diagnosis and care to be effective. False diagnosis and diagnosis at a late stage result into death. Diagnostic capabilities have been greatly improved by recent developments in Artificial Intelligence (AI), especially machine learning and deep learning. However, many AI models, also known as black boxes, are opaque and thus restricted to use in a clinical scenario where interpretability and transparency is important. This paper will look at the application of Explainable AI (XAI) to diagnose leukemia, with a particular focus on how it can be used to provide clear and intelligible explanations of AI-driven decisions. The experimental results prove that the given ensemble model can be useful in classifying the subtypes of leukemia. Explainable AI methods like SHAP and LIME also enable more trust since the insights obtained are transparent and clinically relevant. This demonstrates the possibility of interpretable models being applicable to practice to aid clinical diagnosis. By using XAI techniques on trained model, the potential of XAI to bridge the gap between high-performance AI and clinical applicability is demonstrated. Despite its potential, XAI is faced with several challenges to address, including the need to integrate it into existing clinical workflows, technical complexity, and issues of data protection. At the end of the paper, the importance of developing domain-specific XAI methods and collaborative structures to succeed is outlined.
Open Zoom