Prevalence and Diagnosis of Neurological Disorders Using Different Deep Learning Techniques: A Meta-Analysis

Area of Science:

• Neurology
• Artificial Intelligence
• Medical Informatics

Background:

• Neuropsychiatric and neurological disorders pose significant health risks.
• Deep learning (DL) is a powerful computational technique increasingly applied to complex problems.
• DL has shown success in areas like pattern recognition, drug discovery, and disease diagnosis.

Purpose of the Study:

• To provide a comprehensive review of deep learning techniques for diagnosing eight key neurological and neuropsychiatric disorders.
• To analyze the current landscape, performance trends, and research gaps in DL-based neurological disorder diagnosis.
• To identify future research directions and opportunities for advanced DL models.

Main Methods:

• Systematic review of 136 articles on deep learning applications in diagnosing neurological and neuropsychiatric disorders.
• Analysis of methodologies, frameworks, and performance metrics (accuracy, specificity, sensitivity) of DL techniques.
• Examination of publication trends and morbidity/mortality rates associated with the studied disorders.

Main Results:

• Deep learning techniques are effectively applied to diagnose various neurological conditions.
• Performance metrics of different DL models were analyzed, revealing trends and variations.
• Identified specific disorders like migraine, cerebral palsy, and stroke as areas with significant potential for DL-based diagnosis.

Conclusions:

• Deep learning offers a promising avenue for improving the diagnosis of neurological and neuropsychiatric disorders.
• Further research is warranted to explore advanced DL models, such as Restricted Boltzmann Machines, Deep Boltzmann Machines, and Deep Belief Networks.
• There are clear opportunities to enhance diagnostic capabilities for under-explored conditions using sophisticated DL approaches.