AI in Cardiology

ECG Arrhythmia Detection

import torch.nn as nn

class ECGArrhythmiaClassifier(nn.Module):
    def __init__(self, n_classes=9):
        super().__init__()
        self.conv_layers = nn.Sequential(
            nn.Conv1d(1, 32, 50, stride=5), nn.BatchNorm1d(32), nn.ReLU(),
            nn.MaxPool1d(2), nn.Conv1d(32, 64, 25, stride=2),
            nn.BatchNorm1d(64), nn.ReLU(), nn.MaxPool1d(2),
            nn.Conv1d(64, 128, 10), nn.BatchNorm1d(128), nn.ReLU(),
            nn.AdaptiveAvgPool1d(1))
        self.classifier = nn.Linear(128, n_classes)

    def forward(self, x):
        return self.classifier(self.conv_layers(x).squeeze(-1))

CLASSES = ['Normal', 'AFib', 'AFlutter', 'VTach', 'VFib',
           'LBBB', 'RBBB', 'PVC', 'PAC']

Echocardiogram Analysis

Ejection Fraction

View Classification

Heart Failure Prediction

Evaluation

• AUROC > 0.97 for AFib detection
• MAE < 5% for EF estimation
• Sensitivity > 99% for critical arrhythmias