Early diagnosis of neurodegenerative disorders: A systematic review on multimodal artificial intelligence approaches

Authors

  • Leila Soleimanidoust Islamic Azad University, Department of Medical Engineering, South Tehran Branch
  • Abdalhossein Rezai University of Science and Culture (USC), Faculty of Engineering, Department of Biomedical Engineering, Tehran
  • Hamideh Barghamadi Islamic Azad University, Department of Medical Engineering, South Tehran Branch
  • Iman Ahanian Islamic Azad University, Department of Electrical Engineering, South Tehran Branch

Abstract

Neurodegenerative Disorders (NDDs), including Alzheimer's Disease (AD), Parkinson's Disease (PD), and Multiple Sclerosis (MS), affect the brain and spinal cord, causing motor, cognitive, and sensory impairments. Early detection and diagnosis using Artificial Intelligence (AI) methods, especially multimodal AI, are crucial for accurate treatment and disease management. This systematic review, following PRISMA guidelines, investigates and compares recent advancements in multimodal AI for improving diagnostic accuracy, addressing current trends, and identifying open research gaps. In this paper, total of 72 studies published in the past seven years are selected based on inclusion criteria: studies involving adults diagnosed with NDDs and utilizing AI on multimodal data, such as MRI, cognitive tests, speech, and genetic information. Data are sourced from PubMed, Scopus, IEEE, and Web of Science, encompassing 3,724,703 individuals, including 3,654,166 with Alzheimer’s, 50,465 with Parkinson’s, 20,072 with MS, and 472,938 healthy controls. Findings indicate that multimodal AI approaches, integrating data from various sources, significantly enhance diagnostic accuracy. Key modalities include MRI, cognitive tests, and speech analysis. However, challenges such as limited sample sizes, data heterogeneity, and the lack of standardized protocols persist, highlighting the need for robust validation and ethical frameworks. Despite substantial advancements, further research is needed to address data integration challenges and standardize multimodal approaches. Collaboration among researchers and clinicians will be vital for developing explainable AI models, ultimately improving diagnostic accuracy and treatment outcomes for NDDs.

Additional Files

Published

2026-10-26

Issue

Section

Biomedical Engineering