Diagnosis of Common Headaches Using Hybrid Expert-Based Systems

Area of Science:

• Artificial Intelligence in Medicine
• Medical Informatics
• Computational Intelligence

Background:

• Headaches are prevalent medical complaints with diverse causes and presentations.
• Accurate diagnosis is crucial for effective headache treatment.
• This study addresses the challenge of diagnosing primary and secondary headaches using computational methods.

Purpose of the Study:

• To evaluate the efficacy of artificial intelligence and soft computing techniques for predicting common headache diagnoses.
• To develop and compare intelligent systems for differentiating between various types of headaches.

Main Methods:

• Development of a fuzzy expert system using the Learning-From-Examples (LFE) algorithm with a Mamdani fuzzy inference engine.
• Implementation of Support Vector Machine (SVM) and Multilayer Perceptron (MLP) based classifiers for headache diagnosis.
• Classification of four common headache types: migraine, tension, infection-related, and increased intracranial pressure headaches.

Main Results:

• The fuzzy system, trained with LFE, generated approximately 123 If-Then rules and achieved 85% diagnostic accuracy.
• The MLP-based system achieved 88% accuracy, while the SVM-based system reached 90% accuracy in classifying headache types.
• Performance metrics including accuracy, precision, sensitivity, and specificity were used to evaluate all methods.

Conclusions:

• The LFE algorithm demonstrates superior effectiveness compared to human expert systems for headache diagnosis due to its ability to handle incomplete linguistic rules.
• The developed medical decision support systems based on MLP and SVM show significant utility in recognizing common headaches with overlapping symptoms.
• Artificial intelligence techniques offer a promising approach for improving the accuracy and efficiency of headache diagnosis.