Validation of a new computer program for Minnesota coding

Area of Science:

• Cardiology
• Medical Informatics
• Epidemiology

Background:

• The Minnesota code (MC) is a standard for classifying electrocardiograms (ECGs) in epidemiologic research.
• Manual MC coding is complex, leading to time-consuming and error-prone processes.
• Automating MC coding can improve accuracy and efficiency.

Purpose of the Study:

• To develop and validate a computer program for Minnesota code (MC) classification of ECGs.
• To compare the accuracy of automated MC coding with experienced human readers.
• To assess the program's performance across various MC code categories, including arrhythmias.

Main Methods:

• A software program was developed to adhere strictly to MC measurement procedures and rules.
• A test set of 300 ECGs with diverse codable patterns was used for validation.
• ECGs were independently coded by the program and a human reader, with a consensus procedure establishing the reference code.

Main Results:

• The automated MC program demonstrated sensitivity and specificity comparable to or better than the human reader across all nine main MC categories.
• The program showed particularly strong performance in coding arrhythmias.
• Discrepancies between the program and the reference code often resulted from minor borderline measurement differences and the binary nature of coding criteria.

Conclusions:

• Computerized Minnesota coding is a viable and valuable tool for ECG analysis in epidemiologic studies.
• The developed program serves as an effective alternative or supplement to manual visual coding.
• Automation in ECG classification can significantly reduce errors and enhance the reliability of research findings.