Related Experiment Videos
Predicting ventricular arrhythmia in myocardial ischemia using deep learning
Anna Busatto1,2,3, Jake A Bergquist1,2,3, Tolga Tasdizen1,4
1Scientific Computing and Imaging Institute, University of Utah, Salt Lake City, Utah.
Insights
High-resolution epicardial electrograms can predict the timing of premature ventricular contractions (PVCs) during acute ischemia. Personalized data significantly improves prediction accuracy, paving the way for real-time monitoring tools.
Area of Science:
- Cardiovascular Physiology
- Biomedical Engineering
- Machine Learning in Medicine
Background:
- Myocardial ischemia poses a life-threatening risk due to ventricular arrhythmias.
- Current monitoring methods are reactive, limiting preventive interventions.
Purpose of the Study:
- To determine if high-resolution epicardial electrograms can predict premature ventricular contractions (PVCs) during acute ischemia.
- To quantify data needs for personalized predictive models.
Main Methods:
- Analysis of epicardial sock electrograms from porcine acute ischemia experiments (21 experiments, 2252 PVCs).
- Utilized a 6-layer long short-term memory network with temporal attention to predict time-to-next PVC.
- Evaluated performance using cross-validation and leave-1-experiment-out testing with subject-specific fine-tuning.
Main Results:
- Prediction accuracy (Mean Absolute Error) improved with longer signal context (e.g., 4.73 seconds for 7 beats).
- Subject-specific fine-tuning (10-15% of data) substantially reduced prediction error and improved forecast horizon.
- Excellent calibration (R² > 0.996) was achieved in pooled analysis.
Conclusions:
- Epicardial electrograms provide predictive signatures for PVC timing during ischemia.
- Personalized adaptation of predictive models significantly enhances generalization.
- Findings support the development of real-time predictive monitoring for ventricular arrhythmias.
Background:
Myocardial ischemia can trigger ventricular arrhythmias with life-threatening consequences. Current monitoring is largely reactive, limiting opportunities for preventive intervention.
Objective:
This study aimed to determine whether high-resolution epicardial electrograms contain predictive signatures that enable forecasting the timing of premature ventricular contractions (PVCs) during acute ischemia, and to quantify subject-specific data requirements for effective personalization.
Methods:
We analyzed epicardial sock electrograms (247 electrodes, 1 kHz) from 21 porcine acute ischemia experiments comprising 2252 spontaneous PVCs. Signals were segmented into overlapping sequences of 3, 5, or 7 consecutive non-PVC beats with a continuous target of time-to-next PVC. A 6-layer long short-term memory network (hidden size 128) with temporal attention was trained using mean absolute error (MAE). Performance was evaluated in (A) pooled 80/10/10 cross-validation and (B) leave-1-experiment-out testing with subject-specific fine-tuning using 10% or 15% of held-out data.
Results:
In Paradigm A, MAE decreased with longer context (6.50 seconds for 3 beats, 5.97 seconds for 5 beats, and 4.73 seconds for 7 beats) with excellent calibration (R 2 > 0.996). In Paradigm B, increasing fine-tuning from 10% to 15% reduced mean MAE by 9.6-14.6 seconds and flattened error growth with prediction horizon, improving the fraction of predictions within 30-60 seconds windows.
Conclusion:
Epicardial electrograms support accurate PVC time-to-event forecasting during acute ischemia, and modest subject-specific adaptation substantially improves generalization, motivating development of real-time predictive monitoring tools.
