Gotta love this silliness: "Machine learning overtakes humans in predicting death or heart attack" AUC with only clinical metrics (no imaging etc): .65 AUC with added imaging and machine learning: .82 Conclusion: ML is the savior of all mankind https://www.escardio.org/The-ESC/Press-Office/Press-releases/machine-learning-overtakes-humans-in-predicting-death-or-heart-attack …
-
Show this thread
-
1 reply 0 retweets 0 likesShow this thread
-
- Overtakes humans but doesn’t compare to humans? - Better than existing models but doesn’t compare to them? - Only focusses on AUC and accuracy? - 85 variables in 24 patients with an event? I’ll stop here.
1 reply 2 retweets 8 likes -
Thought you'd like it! I wasn't sure what they meant by 'accuracy', however. PPV?
1 reply 0 retweets 1 like -
Possibly classification accuracy. Note that with only ~3% event fraction, a classifier that predicts no event for everybody already has 97% accuracy
1 reply 1 retweet 7 likes -
Replying to @MaartenvSmeden @GidMK and
On a more serious note though. This seem to be just another hyped paper by uninformed authors. This isn’t helping anyone
1 reply 1 retweet 4 likes -
Replying to @MaartenvSmeden @GidMK and
Referencing to the success of Netflix and Google with millions of users and saying: Hey, you know what, we also almost have a thousand patients! Sigh...
1 reply 1 retweet 2 likes -
Replying to @MaartenvSmeden @GidMK and
Like
@AndrewLBeam has been saying before (in different wording): all this sh*t is drawing attention away from settings where ML/AI can be useful to improve patient care. This hype needs to end1 reply 1 retweet 7 likes -
Replying to @MaartenvSmeden @GidMK and
I agree, though press releases are always worthless. I can't find the study linked in the press release (Shame!), but maybe the actual paper is more sober? When I poured my coffee this morning it was half full, so I'm trying to be charitable.
2 replies 0 retweets 2 likes
Actual paper has yet to be published based on the press release. Looks like this is just being presented at a conference
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.