Machine Learned Cellular Phenotypes in Cardiomyopathy Predict Sudden Death.

RATIONALE: Susceptibility to VT/VF (ventricular tachycardia/fibrillation) is difficult to predict in patients with ischemic cardiomyopathy either by clinical tools or by attempting to translate cellular mechanisms to the bedside. OBJECTIVE: To develop computational phenotypes of patients with ischemic cardiomyopathy, by training then interpreting machine learning of ventricular monophasic action potentials (MAPs) to reveal phenotypes that predict long-term outcomes. METHODS AND RESULTS: We recorded 5706 ventricular MAPs in 42 patients with coronary artery disease and left ventricular ejection fraction ≤40% during steady-state pacing. Patients were randomly allocated to independent training and testing cohorts in a 70:30 ratio, repeated K=10-fold. Support vector machines and convolutional neural networks were trained to 2 end points: (1) sustained VT/VF or (2) mortality at 3 years. Support vector machines provided superior classification. For patient-level predictions, we computed personalized MAP scores as the proportion of MAP beats predicting each end point. Patient-level predictions in independent test cohorts yielded c-statistics of 0.90 for sustained VT/VF (95% CI, 0.76-1.00) and 0.91 for mortality (95% CI, 0.83-1.00) and were the most significant multivariate predictors. Interpreting trained support vector machine revealed MAP morphologies that, using in silico modeling, revealed higher L-type calcium current or sodium-calcium exchanger as predominant phenotypes for VT/VF. CONCLUSIONS: Machine learning of action potential recordings in patients revealed novel phenotypes for long-term outcomes in ischemic cardiomyopathy. Such computational phenotypes provide an approach which may reveal cellular mechanisms for clinical outcomes and could be applied to other conditions.

Machine Learning to Classify Intracardiac Electrical Patterns During Atrial Fibrillation: Machine Learning of Atrial Fibrillation.

BACKGROUND: Advances in ablation for atrial fibrillation (AF) continue to be hindered by ambiguities in mapping, even between experts. We hypothesized that convolutional neural networks (CNN) may enable objective analysis of intracardiac activation in AF, which could be applied clinically if CNN classifications could also be explained. METHODS: We performed panoramic recording of bi-atrial electrical signals in AF. We used the Hilbert-transform to produce 175 000 image grids in 35 patients, labeled for rotational activation by experts who showed consistency but with variability (kappa [κ]=0.79). In each patient, ablation terminated AF. A CNN was developed and trained on 100 000 AF image grids, validated on 25 000 grids, then tested on a separate 50 000 grids. RESULTS: In the separate test cohort (50 000 grids), CNN reproducibly classified AF image grids into those with/without rotational sites with 95.0% accuracy (CI, 94.8%-95.2%). This accuracy exceeded that of support vector machines, traditional linear discriminant, and k-nearest neighbor statistical analyses. To probe the CNN, we applied gradient-weighted class activation mapping which revealed that the decision logic closely mimicked rules used by experts (C statistic 0.96). CONCLUSIONS: CNNs improved the classification of intracardiac AF maps compared with other analyses and agreed with expert evaluation. Novel explainability analyses revealed that the CNN operated using a decision logic similar to rules used by experts, even though these rules were not provided in training. We thus describe a scaleable platform for robust comparisons of complex AF data from multiple systems, which may provide immediate clinical utility to guide ablation. Registration: URL: https://www.clinicaltrials.gov; Unique identifier: NCT02997254. Graphic Abstract: A graphic abstract is available for this article.

Caffe con Troll: Shallow Ideas to Speed Up Deep Learning.

We present Caffe con Troll (CcT), a fully compatible end-to-end version of the popular framework Caffe with rebuilt internals. We built CcT to examine the performance characteristics of training and deploying general-purpose convolutional neural networks across different hardware architectures. We find that, by employing standard batching optimizations for CPU training, we achieve a 4.5× throughput improvement over Caffe on popular networks like CaffeNet. Moreover, with these improvements, the end-to-end training time for CNNs is directly proportional to the FLOPS delivered by the CPU, which enables us to efficiently train hybrid CPU-GPU systems for CNNs.