Introduction
We propose the term “cardiosomnography” (CSG) for any sleep study that could be conducted using only electrocardiography (ECG/EKG) data. In addition to expert-level, five-stage, sleep scoring [1], numerous studies have demonstrated that sleep apnea detection can be reliably performed using only ECG [2, 3, 4].
Our intention is for CSG to take more expert-level sleep studies outside the confines of clinics and laboratories and into realistic settings. By eliminating the need for the most cumbersome equipment and a human scorer, it makes less-expensive, higher-quality studies more widely accessible.
Our intention is for CSG to take more expert-level sleep studies outside the confines of clinics and laboratories and into realistic settings. By eliminating the need for the most cumbersome equipment and a human scorer, it makes less-expensive, higher-quality studies more widely accessible.
Adam M. Jones, Laurent Itti, Bhavin R. Sheth, "Expert-level sleep staging using an electrocardiography-only feed-forward neural network," Computers in Biology and Medicine, 2024, doi: 10.1016/j.compbiomed.2024.108545.
Sleep Staging with CSG
We demonstrate in our recent paper [1] that it is now possible to score sleep at equivalent performance to expert human-scored polysomnography (PSG) using only ECG. The method offers an inexpensive, automated, and convenient alternative for sleep stage classification—further enhanced by a real-time scoring option.
We suggest that everyone interested should read the paper to find out more. (Make sure to also check out the supplemental, as I had too many results for the main text.) However, we provide some summary results here if you would like to read more.
We suggest that everyone interested should read the paper to find out more. (Make sure to also check out the supplemental, as I had too many results for the main text.) However, we provide some summary results here if you would like to read more.
Sleep Score Your Own Data
There are currently three models in the repository that anyone can use to score their own ECG data. The GitHub repository contains everything you need to do the following:
- Prepare your data.
- Sleep score your data.
- Use the benchmark dataset.
- Use the loss function for your own models.
- Replicate everything in the paper’s Methods and Results sections.
Read More
If you’d like to learn more about the following topics, you can read more on the next page:
- The paper's results
- The three available models
- ECG equipment suggestions
- References
Interviews
A list of interviews on the research conducted since the paper was published:
- 2024-08-19: USC Viterbi News (usc.edu)
- 2024-08-05: Healio (healio.com)
- 2024-07-02: UH Newsroom (uh.edu)
- 2024-06-27: NSRR (youtube.com)