DefinePK

Abstract

The core of neurological diseases is the top in the world, as these diseases comprise the Alzheimer disease, Parkinson disease, epilepsy and stroke. And there is an urgent necessity to forecast such conditions in the earliest possible time and with accuracy to treat and administer the disease. It was discovered that the deep neural network (DNN) and machine learning systems could analyze relatively complicated neurological information, neuroimaging features and electrophysiological changes and medical history. However, the complexity of AI models, which is the black-box of most of them, makes them not applicable to using them in clinical practice, since clinicians require a predictive system that is transparent and interpretable and can be expected to rely on it. The field of explainable artificial intelligence (XAI) is inspiring as a promising approach to these issues as it offers a way to interpret model predictions in a human-understandable manner. The current review discusses the existing XAI methods and their use in predicting neurological disorders with a particular focus on their role in improving clinical judgment, patient trust, and regulatory adherence. Besides, we present the issues concerning data quality, generalizability, and ethical issues and the future perspectives of applying XAI to the routine in neurological practice.
Keywords: Explainable AI (XAI), Neurological Disorders, Machine Learning, Deep Learning; Interpretability, Clinical Decision Support, Neuroimaging; Alzheimer’s Disease, Parkinson’s Disease, Epilepsy Prediction.

https://doi.org/10.5281/zenodo.17251411