Intelligent Computer Systems for Multiple Sclerosis Diagnosis: a Systematic Review of Reasoning Techniques and...

Area of Science:

• Medical Informatics
• Artificial Intelligence
• Neurology

Background:

• Intelligent computer systems are increasingly vital for accurate and timely Multiple Sclerosis (MS) diagnosis.
• Physicians benefit from these systems in managing MS.
• This review examines reasoning techniques within these intelligent systems.

Purpose of the Study:

• To review reasoning techniques and methods used in intelligent systems for MS diagnosis.
• To analyze the application and efficiency of various reasoning methods.
• To identify the most efficient and applicable methods for MS diagnosis.

Main Methods:

• A comprehensive literature search was conducted on electronic databases using MeSH vocabulary.
• 85 relevant articles published between 2000 and 2018 were analyzed.
• 30 articles were selected based on inclusion criteria including system scope, reasoning method description, and evaluation.

Main Results:

• Rule-based methods were used in 27% of studies, fuzzy logic in 20%, and artificial neural networks in 18%.
• Other reasoning methods accounted for 35% of applications.
• The average sensitivity, specificity, and accuracy across methods were 0.91, 0.77, and 0.86, respectively.

Conclusions:

• Rule-based, fuzzy logic, and artificial neural network methods are most frequently applied in intelligent MS diagnostic systems.
• Artificial neural networks demonstrated the highest sensitivity (0.97) and accuracy (0.99).
• Fuzzy logic achieved perfect concordance (Kappa rate of 1) with physician diagnoses; combined methods show potential for enhanced efficiency.