AUTOMATED VENTRICULAR ECTOPIC BEAT CLASSIFICATION
Techniques for classifying heartbeats using patient electrocardiogram (ECG) data are described. ECG data is received, including waveform data and time interval data relating to a plurality of heartbeats for the patient. A convolutional neural network in a first path of a machine learning architecture generates a first plurality of output values by analyzing the waveform data. A fully-connected neural network in a second path of the machine learning architecture generates a second plurality of output values by analyzing the time interval data. The plurality of heartbeats in the ECG data are classified by concatenating the first plurality of output values and the second plurality of output values using the machine learning architecture.
1 . A computer-implemented method for classifying heartbeats using patient electrocardiogram (ECG) data, which comprises unstructured waveform data and structured time interval data, the method comprising:
providing the unstructured waveform data to a first neural network in a first path of a machine learning architecture;
generating a first set of output values by using the first neural network to analyze the unstructured waveform data;
providing structured time interval data to a second neural network in a second path of the machine learning architecture;
generating a second set of output values by using the first neural network to analyze the structured time interval data; and
classifying the heartbeats based on the first set of output values and the second set of output values, wherein the first neural network comprises a different type of neural network than the second neural network.
2 . The computer-implemented method of claim 1 , wherein the first set of output values and the second set of output values are concatenated using a fully connected layer in the machine learning architecture.
3 . The computer-implemented method of claim 2 , wherein the fully connected layer outputs a probability for each heartbeat classification.
4 . The computer-implemented method of claim 2 , wherein the classifying comprises: verifying, using the machine learning architecture, that the heartbeat comprises a beat, based on concatenation of the first set of output values and the second set of output values.
5 . The computer-implemented method of claim 1 , wherein the first neural network comprises a convolutional neural network.
6 . The computer-implemented method of claim 5 , wherein the second neural network comprises a fully-connected neural network.
7 . The computer-implemented method of claim 6 , wherein the second path does not comprise a convolution layer.
8 . The computer implemented method of claim 1 , wherein the first neural network does not process the structured time interval data, wherein the second neural network does not process the unstructured waveform data.
9 . The computer-implemented method of claim 1 , wherein the structured time interval data comprises R-R interval data.
10 . The computer-implemented method of claim 1 , wherein the first neural network analyzes separate groups of heartbeat trains at a time.
11 . A system for classifying heartbeats using patient electrocardiogram (ECG) data, which comprises unstructured waveform data and structured time interval data, the system comprising:
one or more servers, each including one or more processors and computer-readable memory, wherein instructions are stored on the memory and cause the one or more servers to perform an operation comprising:
providing the unstructured waveform data to a first neural network in a first path of a machine learning architecture;
generating a first set of output values by using the first neural network to analyze the unstructured waveform data;
providing structured time interval data to a second neural network in a second path of the machine learning architecture;
generating a second set of output values by using the first neural network to analyze the structured time interval data; and
classifying the heartbeats based on the first set of output values and the second set of output values, wherein the first neural network comprises a different type of neural network than the second neural network.
12 . The system of claim 11 , wherein the first set of output values and the second set of output values are concatenated using a fully connected layer in the machine learning architecture.
13 . The system of claim 12 , wherein the fully connected layer outputs a probability for each heartbeat classification.
14 . The system of claim 12 , wherein the classifying comprises: verifying, using the machine learning architecture, that the heartbeat comprises a beat, based on concatenation of the first set of output values and the second set of output values.
15 . The system of claim 11 , wherein the first neural network comprises a convolutional neural network.
16 . The system of claim 15 , wherein the second neural network comprises a fully-connected neural network.
17 . The system of claim 16 , wherein the second path does not comprise a convolution layer.
18 . The system of claim 11 , wherein the first neural network does not process the structured time interval data, wherein the second neural network does not process the unstructured waveform data.
19 . The system of claim 11 , wherein the structured time interval data comprises R-R interval data.
20 . The system of claim 11 , wherein the first neural network analyzes separate groups of heartbeat trains at a time.