Hacker News Reader

This was a collaboratively written review paper, published on github with submissions encouraged in the form of pull requests, with co-author credit given to anyone whose contributions met the ICJME standards of authorship. Very cool.

I'm a bit disappointed as I only found out about this a month or so ago. I would have loved to contribute to it, if only because of how the paper was written. Just fantastic.

View on HN

I found this comment at least as interesting as the paper, but I can't find the github repo for it, do you have a link?

Oops, it's a different paper. When I was googling Deep Patient a few months ago I must have got confused because they mention the Deep Patient paper in their issue tracker:

https://github.com/greenelab/deep-review/issues/63

Same here, especially since I am working on the same problem and in fact also plan to try out denoising autoencoders to reduce the sparse EMR data into more learnable features. I am also using NLP on the medical notes to generate additional features. It is good to see promising results from these approaches. I will read this paper closely tomorrow but from what I have skimmed it looks well written.

Reply on news.ycombinator.com