Comparison of discriminant analysis and probabilistic expert system in VCG data classification

Area of Science:

• Cardiology
• Medical Informatics
• Artificial Intelligence in Medicine

Background:

• Previous research indicated the suitability of probabilistic expert systems for electrocardiologic data interpretation.
• Vectorcardiography (VCG) data analysis presents challenges in accurate patient classification.

Purpose of the Study:

• To compare the classification performance of a probabilistic expert system against stepwise discriminant analysis.
• To evaluate these methods on VCG data from diverse patient groups, including healthy individuals and those with myocardial infarction or angina pectoris.

Main Methods:

• Application of a probabilistic expert system and stepwise discriminant analysis to VCG data.
• Utilized Frank's lead system for VCG data acquisition.
• Employed the leaving-one-out technique for robust classification validation.
• Investigated five patient groups: healthy subjects, angina pectoris, posterior myocardial infarction, anterior myocardial infarction, and anteroseptal myocardial infarction.

Main Results:

• The probabilistic expert system achieved superior classification results compared to stepwise discriminant analysis across all five patient groups.
• Demonstrated significant differences in classification accuracy favoring the expert system for VCG data interpretation.

Conclusions:

• Probabilistic expert systems offer a more effective approach for classifying VCG data than traditional statistical methods like stepwise discriminant analysis.
• The findings support the continued development and application of AI-driven systems in cardiac diagnostics.