PyApnea v0.0.4 - Apnea to Datascience

[jlaumonier]

Hi everyone,

I am pleased to announce the development of pyapnea, a python library designed to help data-scientists working with apnea related data files. The preliminary version 0.0.4 can read session files from Oscar. It transforms data into Pandas Dataframes for further analysis or visualization purposes.

Please note that this is a very early release, which means it's still quite limited in terms of functionalities compared to what we plan to achieve with future updates and enhancements. Nonetheless, if you have any suggestions, comments, requests for new features or simply want to contribute to the project, feel free to do so!

Pypi: <https://pypi.org/project/pyapnea/>
GitHub: <https://github.com/iid-ulaval/pyapnea>

Have a nice and pleasant day,

Julien Laumonier

[chemmkl]

Excellent work, Julien! Do you have some API usage examples that you can share?

Thanks!

[jlaumonier]

Excellent work, Julien! Do you have some API usage examples that you can share?

Thanks!

Thank you for the comment. I forgot to add examples but i have created an issue and will do it soon.

Basically it's easy, load_session() takes a string path to a .001 file and return a data structure. event_data_to_dataframe() takes that data structure and return a pandas dataframe that can be used with any python datascience library.

Julien

[EddyDee]

Julien
It seems like the statistical measure "Ventilatory Burden" proposed recently by the group at Mount Sinai Hospital would be a nice application for this library. See:
https://www.apneaboard.com/forums/Thread...den-vs-AHI
https://www.apneaboard.com/forums/Thread...for-future

What do you think?

[jstro84]

Oh this is so neat! I can't wait to try it out and analyze my data with it! Are you still actively developing it? I would be interested in helping

[jlaumonier]

Oh this is so neat! I can't wait to try it out and analyze my data with it! Are you still actively developing it? I would be interested in helping

I am   However, this fall I had too much work to, i took a theoretical course in machine learning and temporal series to understand what could be done with these type of data and i was close to have a burnout. So sorry, I didn't work on PyApnea as i expected. But, in january, i should have more time.

If you are interessed in helping, do not hesitate to discuss with me in private message. If you have some ideas to  improve this library and have some programming skill, it would be very helpful. I have some other code that could be implemented in pyapnea (like pytorch data management and an LSTM algorithm) but i am not sure if they are generic enough to be implemented in PyApnea.

Have a nice day

Julien

[jlaumonier]

Julien
It seems like the statistical measure "Ventilatory Burden" proposed recently by the group at Mount Sinai Hospital would be a nice application for this library. See:
[...]

What do you think?

Oh ! This seems nice. If they have a new measure of sleep apnea, better than AHI for some aspect, this could change the way that the apnea could be cured. I will check in detail this article in january. I hope they gave details of their algorithm in their article.

Thanks,

Have a nice day

Julien

[EddyDee]

He has MATLAB code at https://github.com/aparek/breathmetrics, but I'm not sure if that covers everything in the paper.

[jlaumonier]

He has MATLAB code at https://github.com/aparek/breathmetrics, but I'm not sure if that covers everything in the paper.

Nice ! I will definitely try to implement these metrics in python to use them in my research. At the moment using the raw signal is not as good as i expected so i thinl I need to extract features from the signal.

The only bad part is that i am absolutely not an expert of Matlab but i will try.

Julien

[EddyDee]

It's definitely worth reading the paper too