Development of a Knowledge-based Clinical Decision Support System for Multiple Sclerosis Diagnosis

Area of Science:

• Medical Informatics
• Neurology
• Artificial Intelligence in Healthcare

Background:

• Diagnosing multiple sclerosis (MS) presents challenges due to its complex and varied presentation, often mimicking other neurological conditions.
• Accurate and timely diagnosis is crucial for effective management of MS, particularly the relapsing-remitting phenotype.

Purpose of the Study:

• To design and develop a clinical decision support system (CDSS) to assist physicians in diagnosing multiple sclerosis (MS).
• To evaluate the diagnostic performance, applicability, and user satisfaction of the developed CDSS for MS diagnosis.

Main Methods:

• A four-stage development process: requirement analysis, system design, software development (using C# and Visual Studio), and database development (SQL Server).
• Object-oriented conceptual model designed with Rational Rose.
• Evaluation involved analyzing 130 patient records (aged 20-40, 2017-2019), using the Nilsson standard questionnaire, and statistical analysis with SPSS.

Main Results:

• The CDSS achieved high diagnostic performance for MS: sensitivity of 1, specificity of 0.97, accuracy of 0.99, and an ROC curve area of 0.98.
• Excellent agreement between CDSS and physician diagnoses (kappa coefficient κ = 0.98).
• High user satisfaction scores for ease of learning (98.33%), memorability (96.65%), and overall satisfaction (96.9%).

Conclusions:

• The developed CDSS is effective and applicable for aiding physicians in the accurate and timely diagnosis of multiple sclerosis.
• The rule-based CDSS demonstrates significant potential to improve diagnostic workflows in neurology.
• Neurologist confirmation supports the clinical utility and reliability of the CDSS for MS diagnosis.