1,2,3Department of Artificial Intelligence and Machine Learning, Karunya Institute of Technology and Sciences, Coimbatore, India
1Email: arend@karunya.edu.in
2Email: amandadsouza@karunya.edu.in
3Email: christinarvs@karunya.edu
Hepatic conditions constitute a significant global health challenge due to their miscellaneous clinical instantiations and the limitations of conventional individual workflows. Although artificial intelligence has shown considerable promise in automating liver complaint discovery, the lack of interpretability in numerous deep learning models restricts their clinical abandonment. This study presents a new resolvable artificial intelligence (XAI) frame designed to deliver transparent, dependable, and clinically interpretable decision support for liver complaint opinion. The proposed frame employs deep neural network models trained on hepatic medical imaging data, including ultrasound, computed tomography, and magnetic resonance images, to capture complex pathological patterns associated with a wide spectrum of liver diseases. To overcome the nebulosity of traditional black-box models, the system integrates post-hoc explainability ways that induce spatial and point-grounded attributions, pressing diagnostically applicable regions and variables that impact model prognostications. These explanations are structured to align with established hepatological assessment practices, easing clinician understanding and confirmation. Likewise, the frame incorporates multimodal clinical information, such as liver function test parameters, patient demographics, and medical history, enabling a comprehensive and environment-apprehensive individual process.
Keywords: Resolvable Artificial Intelligence, Liver Disease opinion, Hepatic Imaging, Clinical Decision Support Systems, Deep Learning, Model translucency, Multimodal Data Fusion.
How to cite this article: D'Souza A, D'Souza A, Magneta SC. Decoding Hepatic Pathologies: A Novel Explainable AI Framework for Transparent Liver Disease Decision Support. Int J Drug Deliv Technol. 2026;16(6s): 1006-1011; DOI: 10.25258/ijddt.16.6s.131
Source of support: None
Conflict of interest: None