Citation

Al-bakri, Fatima Hasan and Wan Bejuri, Wan Mohd Yaakob and Al-Andoli, Mohammed Nasser and Raja Ikram, Raja Rina and Khor, Hui Min and Sholva, Yus and Ariffin, Umi Kalsom and Mohd Yaacob, Noorayisahbe and Abas, Zuraida Abal and Zainal Abidin, Zaheera and Asmai, Siti Azirah and Ahmad, Asmala and Abdul Rahman, Ahmad Fadzli Nizam and Rahmalan, Hidayah and Abd Samad, Md Fahmi (2025) A Feature-Augmented Explainable Artificial Intelligence Model for Diagnosing Alzheimer’s Disease from Multimodal Clinical and Neuroimaging Data. Diagnostics, 15 (16). p. 2060. ISSN 2075-4418

Text A Feature-Augmented Explainable Artificial Intelligence Model for Diagnosing Alzheimer’s Disease from Multimodal Clinical and Neuroimaging Data.pdf - Published Version Restricted to Repository staff only Download (960kB)

Abstract

Background/Objectives: This study presents a survey-based evaluation of an explainable AI (Feature-Augmented) approach, which was designed to support the diagnosis of Alzheimer’s disease (AD) by integrating clinical data, MMSE scores, and MRI scans. The approach combines rule-based reasoning and example-based visualization to improve the explainability of AI-generated decisions. Methods: Five doctors participated in the survey: two with 6 to 10 years of experience and three with more than 10 years of experience in the medical field and expertise in AD. The participants evaluated different AI outputs, including clinical feature-based interpretations, MRI-based visual heat maps, and a combined interpretation approach. Results: The model achieved a 100% trust score, with 20% of the participants reporting full trust and 80% expressing conditional trust, understanding the diagnosis but seeking further clarification. Overall, the participants reported that the integrated explanation format improved their understanding of the model decisions and enhanced their confidence in using AI-assisted diagnosis. Conclusions: To our knowledge, this paper is the first to gather the views of medical experts to evaluate the explainability of an AI decision-making model when diagnosing AD. These preliminary findings suggest that explainability plays a key role in building trust and ease of use of AI tools in clinical settings, especially when used by experienced clinicians to support complex diagnoses, such as AD.

Item Type:	Article
Uncontrolled Keywords:	Alzheimer
Subjects:	Q Science > Q Science (General) > Q300-390 Cybernetics R Medicine > RC Internal medicine > RC71-78.7 Examination. Diagnosis
Divisions:	Faculty of Computing and Informatics (FCI)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	29 Sep 2025 09:11
Last Modified:	04 Oct 2025 12:34
URII:	http://shdl.mmu.edu.my/id/eprint/14509

Downloads

Edit (login required)