{ "id": 467129, "title": "Resources for getting started with Heart Data - ECG/EKG", "url": "/competitions/hms-harmful-brain-activity-classification/discussion/467129", "author_name": "", "post_date": "2024-01-11T08:44:20.355914500Z", "votes": 27, "comment_count": 3, "views": 0, "content": "

If you haven't read my previous post on getting started with brain data follow: Resources for getting started with Brain Data - EEG

\n

The main features of ECG are the QRS complexes, R peaks, and R-R intervals
\n link

\n", "messages": [ { "id": "2596662", "postDate": "01/11/2024 08:44:20", "content": "

If you haven't read my previous post on getting started with brain data follow: Resources for getting started with Brain Data - EEG

\n

The main features of ECG are the QRS complexes, R peaks, and R-R intervals
\n link

\n", "rawMarkdown": "**If you haven't read my previous post on getting started with brain data follow**: [Resources for getting started with Brain Data - EEG](https://www.kaggle.com/competitions/hms-harmful-brain-activity-classification/discussion/466768)\n\nThe main features of ECG are the QRS complexes, R peaks, and R-R intervals\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2030387%2F1fdc5942d17ce7b15801884d74241d66%2FScreenshot%202024-01-11%20093911.png?generation=1704962385485913&alt=media) [link](https://www.researchgate.net/publication/372498054_Memory_Classifiers_for_Robust_ECG_Classification_against_Physiological_Noise)\n\n- To detect these features, there exist multiple Python and Matlab packages, follow [this ](https://www.samproell.io/posts/signal/ecg-library-comparison/)blogpost \n- A good paper highlighting R peak detection with decision trees https://www.mdpi.com/1424-8220/21/19/6682\n- Another really good resource describing the characteristics and the procedure of extracting most of the useful features from ECG data: https://asifr.com/machine-learning-digital-health/chapter03/beat-detection\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2030387%2F8de171d5fde4afbadc1691ababdde3d8%2Fecg-morphology.png?generation=1704962638836359&alt=media)", "votes": null }, { "id": "2637788", "postDate": "02/05/2024 21:36:14", "content": "

Hi @iworeushankaonce, Thanks for the amazing resources here -- I'm learning a lot about EKG. Your blog post you cite is a great starting place for me. As the blog post says, it might not be important for the peaks to be exactly correct, but here's another approach here that may improve the result if deemed necessary. I've always thought that calibrating the EKG to the EEG could remove some noise like here

", "rawMarkdown": "Hi @iworeushankaonce, Thanks for the amazing resources here -- I'm learning a lot about EKG. Your blog post you cite is a great starting place for me. As the blog post says, it might not be important for the peaks to be exactly correct, but here's another approach [here](https://www.nature.com/articles/s41598-022-19495-9) that *may* improve the result if deemed necessary. I've always thought that calibrating the EKG to the EEG *could* remove some noise like [here](https://pubmed.ncbi.nlm.nih.gov/37149738/)", "votes": null }, { "id": "2638981", "postDate": "02/06/2024 15:48:36", "content": "

haha, you would be surprised but the paper \"Removing the cardiac field artifact from the EEG using neural network regression\" is from our institute 😅

\n

Thanks for the resources!

", "rawMarkdown": "haha, you would be surprised but the paper \"Removing the cardiac field artifact from the EEG using neural network regression\" is from our institute 😅\n\nThanks for the resources!", "votes": null }, { "id": "2638987", "postDate": "02/06/2024 15:55:37", "content": "

The world of science is so small! Have you tried wrangling the EKG data in your models yet? From my end, it's pretty preliminary, but the EKG values are some of the messiest data I've ever seen. Huge swings in orders of magnitude or the mV and crazy frequency distributions from EKG to EKG

", "rawMarkdown": "The world of science is so small! Have you tried wrangling the EKG data in your models yet? From my end, it's pretty preliminary, but the EKG values are some of the messiest data I've ever seen. Huge swings in orders of magnitude or the mV and crazy frequency distributions from EKG to EKG", "votes": null } ], "comments": [ { "id": 2637788, "author_name": "m000sey", "author_url": "", "post_date": "02/05/2024 21:36:14", "content": "

Hi @iworeushankaonce, Thanks for the amazing resources here -- I'm learning a lot about EKG. Your blog post you cite is a great starting place for me. As the blog post says, it might not be important for the peaks to be exactly correct, but here's another approach here that may improve the result if deemed necessary. I've always thought that calibrating the EKG to the EEG could remove some noise like here

", "votes": null, "replies": [ { "id": 2638981, "author_name": "iworeushankaonce", "author_url": "", "post_date": "02/06/2024 15:48:36", "content": "

haha, you would be surprised but the paper \"Removing the cardiac field artifact from the EEG using neural network regression\" is from our institute 😅

\n

Thanks for the resources!

", "votes": null, "replies": [ { "id": 2638987, "author_name": "m000sey", "author_url": "", "post_date": "02/06/2024 15:55:37", "content": "

The world of science is so small! Have you tried wrangling the EKG data in your models yet? From my end, it's pretty preliminary, but the EKG values are some of the messiest data I've ever seen. Huge swings in orders of magnitude or the mV and crazy frequency distributions from EKG to EKG

", "votes": null, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "2596662": "**If you haven't read my previous post on getting started with brain data follow**: [Resources for getting started with Brain Data - EEG](https://www.kaggle.com/competitions/hms-harmful-brain-activity-classification/discussion/466768)\n\nThe main features of ECG are the QRS complexes, R peaks, and R-R intervals\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2030387%2F1fdc5942d17ce7b15801884d74241d66%2FScreenshot%202024-01-11%20093911.png?generation=1704962385485913&alt=media) [link](https://www.researchgate.net/publication/372498054_Memory_Classifiers_for_Robust_ECG_Classification_against_Physiological_Noise)\n\n- To detect these features, there exist multiple Python and Matlab packages, follow [this ](https://www.samproell.io/posts/signal/ecg-library-comparison/)blogpost \n- A good paper highlighting R peak detection with decision trees https://www.mdpi.com/1424-8220/21/19/6682\n- Another really good resource describing the characteristics and the procedure of extracting most of the useful features from ECG data: https://asifr.com/machine-learning-digital-health/chapter03/beat-detection\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2030387%2F8de171d5fde4afbadc1691ababdde3d8%2Fecg-morphology.png?generation=1704962638836359&alt=media)", "2637788": "Hi @iworeushankaonce, Thanks for the amazing resources here -- I'm learning a lot about EKG. Your blog post you cite is a great starting place for me. As the blog post says, it might not be important for the peaks to be exactly correct, but here's another approach [here](https://www.nature.com/articles/s41598-022-19495-9) that *may* improve the result if deemed necessary. I've always thought that calibrating the EKG to the EEG *could* remove some noise like [here](https://pubmed.ncbi.nlm.nih.gov/37149738/)", "2638981": "haha, you would be surprised but the paper \"Removing the cardiac field artifact from the EEG using neural network regression\" is from our institute 😅\n\nThanks for the resources!", "2638987": "The world of science is so small! Have you tried wrangling the EKG data in your models yet? From my end, it's pretty preliminary, but the EKG values are some of the messiest data I've ever seen. Huge swings in orders of magnitude or the mV and crazy frequency distributions from EKG to EKG" }, "source": "meta" }