A decision support system and rule-based algorithm to augment the human interpretation of the 12-lead...

Area of Science:

• Cardiology
• Medical Informatics
• Artificial Intelligence in Medicine

Background:

• The 12-lead Electrocardiogram (ECG) is a cornerstone of cardiac diagnostics but interpretation is complex, leading to cognitive workload and potential errors.
• Existing ECG interpretation methods require significant expertise, increasing the risk of missed diagnoses and suboptimal patient treatment.
• A novel computerised decision support system, Interactive Progressive based Interpretation (IPI), was developed to augment human interpretation.

Purpose of the Study:

• To enhance the accuracy of 12-lead ECG interpretation.
• To reduce the incidence of missed co-abnormalities in cardiac diagnostics.
• To evaluate the efficacy of a semi-automatic interpretation model using a digital assistant for ECG analysis.

Main Methods:

• Development of the Differential Diagnoses Algorithm (DDA) using web technologies and JSON for diagnostic criteria.
• Implementation of a rule-based reasoning algorithm within the IPI system to suggest diagnoses based on interpreter annotations.
• A counterbalanced trial comparing conventional ECG interpretation with the IPI+DDA approach.

Main Results:

• The IPI+DDA approach demonstrated an 8.7% improvement in diagnostic accuracy, though not statistically significant (p=0.1852).
• The system correctly suggested interpretations more often than human interpreters in 7 out of 10 cases.
• Human interpretation accuracy improved to 70% when augmented with seven computerised diagnostic suggestions.

Conclusions:

• The decision support tool increased correct ECG interpretations and the DDA algorithm outperformed human interpretation in suggesting correct diagnoses.
• Computerised diagnostic suggestions from the IPI+DDA system augmented human decision-making in ECG interpretation.
• Larger sample sizes may be required to achieve statistical significance in future studies.