Computerized serial comparison of electrocardiograms: program performance in myocardial infarction.

Area of Science:

• Cardiology
• Medical Informatics
• Artificial Intelligence in Medicine

Background:

• Serial electrocardiograms (ECGs) are crucial for monitoring cardiac changes.
• Detecting subtle diagnostic shifts between ECGs can be challenging for clinicians.
• Automated analysis tools aim to enhance the efficiency and accuracy of ECG interpretation.

Purpose of the Study:

• To evaluate the accuracy and utility of a serial ECG comparison program.
• To compare the program's diagnostic results against expert cardiologist assessments.
• To determine if prior cardiologist editing of previous ECGs impacts comparison accuracy.

Main Methods:

• Utilized a database of ECGs from patients with symptoms suggestive of acute myocardial infarction.
• Compared the diagnostic outputs of the automated serial comparison program with those of an expert cardiologist.
• Assessed agreement levels under two conditions: comparing current ECGs with prior edited ECGs, and analyzing current ECGs in isolation.

Main Results:

• The serial comparison program demonstrated higher agreement with the expert cardiologist when comparing a current ECG with a previously edited ECG.
• Analysis of current ECGs in isolation showed a lower level of agreement compared to the serial, edited comparison.
• The program effectively identified significant measurement differences across diagnostic categories.

Conclusions:

• Cardiologist-edited serial ECG comparison significantly enhances diagnostic accuracy.
• Automated ECG analysis programs are more effective when leveraging prior clinical interpretations.
• This technology shows promise for improving the detection of acute myocardial infarction and other cardiac conditions.