Fuzzy cognitive map based approach for determining the risk of ischemic stroke

Area of Science:

• Neurology
• Artificial Intelligence
• Computational Medicine

Background:

• Stroke is a leading global cause of mortality, with diagnosis complicated by numerous risk factors.
• Effective and immediate stroke diagnosis is crucial for timely treatment and improved patient outcomes.
• Intelligent systems offer potential for enhancing stroke risk assessment and diagnosis.

Purpose of the Study:

• To propose and evaluate a soft computing method, fuzzy cognitive mapping (FCM), for diagnosing ischemic stroke risk.
• To assess the performance of the FCM model against established machine learning algorithms.
• To provide a reliable tool for identifying individuals at high risk of ischemic stroke.

Main Methods:

• Fuzzy cognitive mapping (FCM) was employed for ischemic stroke risk diagnosis.
• Non-linear Hebbian learning was utilized for training the fuzzy cognitive maps.
• The system's risk rate determination was based on expert neurologist opinions.
• Model performance was validated using 110 real cases with 10-fold cross-validation.

Main Results:

• The proposed FCM system demonstrated superior performance compared to Support Vector Machine (SVM) and K-Nearest Neighbors (KNN).
• The system achieved a high accuracy rate of (93.6 ± 4.5)%.
• The results indicate the effectiveness of FCM in stroke risk diagnosis.

Conclusions:

• Fuzzy cognitive mapping, trained with non-linear Hebbian learning, is a highly accurate method for ischemic stroke risk diagnosis.
• The developed intelligent system shows significant potential for clinical application in stroke prevention and management.
• The study highlights the utility of soft computing approaches in complex medical diagnostic challenges.