Using machine learning techniques Stanford University researchers reported developing an algorithm for identifying cardiac arrhythmias that performs as well or better than cardiologists. Training the model, as usual, was the big hurdle. The researchers used a 34-layer convolutional neural network (CNN) to train a model able to distinguish 14 types of arrhythmias.
The new work is from Stanford’s Machine Learning Group, which is led by Andrew Ng, and was reported last week in an arXiv paper (Cardiologist-Level Arrhythmia Detection with Convolutional Neural Networks) and in an article on the Stanford site. The study and result is more evidence of machine learning’s rapid spread into diverse applications.
“Given that more than 300 million ECGs are recorded annually, high-accuracy diagnosis from ECG can save expert clinicians and cardiologists considerable time and decrease the number of misdiagnoses. Furthermore, we hope that this technology coupled with low-cost ECG devices enables more widespread use of the ECG as a diagnostic tool in places where access to a cardiologist is difficult,” write the paper’s authors.
The effective use of CNNs and a large database were instrumental to the project’s success. “We build a dataset (30,000 unique patients) with more than 500 times the number of unique patients than previously studied corpora. On this dataset, we train a 34-layer convolutional neural network which maps a sequence of ECG samples to a sequence of rhythm classes. Committees of board-certified cardiologists annotate a gold standard test set on which we compare the performance of our model to that of 6 other individual cardiologists. We exceed the average cardiologist performance in both recall (sensitivity) and precision (positive predictive value),” report the researchers.
Stanford worked with iRhythm, a provider of cardiac monitoring systems, on the study. Data were collected from iRhythm’s wearable ECG monitor. Patients wear a small chest patch for two weeks and carry out their normal day-to-day activities while the device records each heartbeat for analysis. The group took approximately 30,000, 30-second clips from various patients that represented a variety of arrhythmias.
As quoted in the Stanford article written by Taylor Kubota, “The differences in the heartbeat signal can be very subtle but have massive impact in how you choose to tackle these detections. For example, two forms of the arrhythmia known as second-degree atrioventricular block look very similar, but one requires no treatment while the other requires immediate attention,” said Pranav Rajpurkar, a graduate student and co-lead author of the paper.
To test accuracy of the algorithm, “the researchers gave a group of three expert cardiologists 300 undiagnosed clips and asked them to reach a consensus about any arrhythmias present in the recordings. Working with these annotated clips, the algorithm could then predict how those cardiologists would label every second of other ECGs with which it was presented, in essence, giving a diagnosis.”
Link to paper: https://arxiv.org/pdf/1707.01836.pdf
Link to Stanford article, written by Taylor Kubota: http://news.stanford.edu/2017/07/06/algorithm-diagnoses-heart-arrhythmias-cardiologist-level-accuracy/