ECG-only expert-level sleep staging.
We just updated the U.S. census benchmark dataset (introduced and described in the original paper). It was brought to our attention that 16 of the recordings had annotations that were ontologically mismatched for the goal of full sleep stage classification. Therefore, we replaced them with age- and sex-matched recordings suitable for full sleep stage classification. The updated dataset is on the…
Now that the 12 month embargo period has passed, I can share the “accepted manuscript” version of the paper. This is identical, except in formatting, to the “published journal article” version of the paper. accepted manuscript version FYI: The article sharing policies for Elsevier are here: Article Sharing.
I’ve had a few discussions in the last month on using the model to score data during the daytime for both a follow-up scientific study as well as personal research. Paper’s dataset As I detailed in the Methods of the paper, the data came from 5 different sleep datasets, with the only “time” requirement being that the recording was from 5 to 15hrs long. As a consequence, almost all of the data was…
Below is the history of and my future plans for cardiosomnography (CSG). I will try to keep this post evergreen. 2009-2013 My self-initiated research began in the fall of 2009, and over the next 4 years I developed several devices, software, and an iPhone app. 2009 I started meditating and immediately began looking for an objective way to track my progress. In December, I built my first device for…
Since I had published the diagram of “roughly” the ECG scaling that should be used for the network input, I have wanted to quantify how much that scaling matters to the network. One thing that was very obvious by looking at the data from the different NSRR studies was that the ECG amplitudes were all over the place, even when using the same equipment. So, I needed to normalize these amplitudes as…
As I mentioned on the ECG Equipment section of the Requirements page, I am currently only aware of two commercial ECG sensors that might work for all-night ECG recordings. I can now begin testing these two, as Movesense recently sent me a Movesense MD sensor (thank you, Movesense) and I purchased a Polar H10 sensor. Furthermore, since there are currently no (100%) free apps that allow users to…
This interview was conducted by the University of Southern California Viterbi News. Interview at USC Viterbi News (usc.edu)
I received some requests to provide a version of the model that does not require demographic information (age and sex). Therefore, I trained a model where that information is not required (and, indeed, is ignored if provided). The model works the same way as the primary model, except it ignores any demographic information provided. The Cohen’s kappa for this model on the testing set is 0.718,…
This interview was conducted by Healio. Interview at Healio (healio.com)
This interview was conducted by the University of Houston Newsroom. Interview at UH Newsroom (uh.edu)