Neurodegenerative disorders: A Holistic study of the explainable artificial intelligence applications

作者：Hongyuan Wang, Shiva Toumaj, Arash Heidari, Alireza Souri, Nima Jafari, Yiping Jiang · 发表于：Engineering Applications of Artificial Intelligence · 年份：2025 · DOI：10.1016/j.engappai.2025.110752 · 被引用次数：26 · 研究领域：Machine Learning in Healthcare、Explainable Artificial Intelligence (XAI)

Neuro Degenerative Disorders (NDDs) involve progressive nerve cell loss, impacting functions like sensation, movement, memory, and cognition, posing life-threatening risks. Despite extensive research, viable therapies remain elusive due to complex pathophysiology. Artificial Intelligence (AI), particularly Machine Learning (ML) and Deep Learning (DL), shows promise in NDD diagnosis and treatment by leveraging vast datasets for accurate predictions. However, because AI models are “black boxes,” explainable AI (XAI) had to be created to make sure that physicians and patients would trust and accept it. Early detection is critical to stop degeneration and make things better for patients. Many in-depth studies on XAI are designed explicitly for NDDs. Existing research does not constantly look at how to interpret NDDs, how to evaluate them, or how to keep them safe. This paper fills in these gaps by looking at and grouping XAI methods for different NDDs, to make them easier to understand and use in medical settings. In this paper, we look at the interpretability methods used in various NDD studies. The methods are split into five groups based on the conditions they are used to treat: Frontotemporal Dementia (FTD), Multiple Sclerosis (MS), Amyotrophic Lateral Sclerosis (ALS), and Alzheimer's Disease (AD). It organizes XAI methods into groups and talks about their pros, cons, and clinical importance. The study also finds some important research gaps. For example, it says that there a...