johannes_ne

Sorry, my point was not that you had taken the idea or the images (I also used ISIC).

I just considered that the language model (Gemini) may have been especially effective at coding this specific app idea, sine my old app (which is on GitHub) was probably in the data it was trained on.

I made a quite similar app 7 years ago. https://melanoma.jenevoldsen.com/

May have been in the training data.

The same author has a great series of tutorials for making looping animations, where the animation appears longer than the length of the loop. It also introduces Perlin noise, which I've found quite useful in a few unrelated projects.

https://bleuje.com/tutorials/

Here is an example of a 10-minute movie in a 6.3-second looping GIF. https://bsky.app/profile/johsenevoldsen.bsky.social/post/3lm...

Indeed, but that problem is just so much more complex.

Thank you for all the kind responses.

I also want to make a similar article, where I calculate an ECG for the simulation, and then make and explain the changes necessary to make the ECG look realistic. A main challenge will be that the depolarization has to happen very fast relative to the repolarization, which may be computationally difficult for a cell-based simulation.

Thank you all for the interest in the article and the kind feedback.

I'm a medical doctor with an interest in engineering and coding. My PhD was quite focused on signal analysis and coding, and my supervisor is a medical engineer.

There is this fork which has a few different languages and some other cool features. I believe the non-english quotes are mostly machine translations of the quotes in my repo. https://literatureclock.netlify.app/?locale=pt-BR

I made https://literature-clock.jenevoldsen.com/

It is based upon a Kindle project [1] and initially I just used the quote library from that project (which is based on a crowd-sourced collection of quotes by The Guardian [2]).

Later, a lot of quotes have been added by kind strangers through GitHub issues and pull-requests [3].

[1] https://www.instructables.com/Literary-Clock-Made-From-E-rea...

[2] https://www.theguardian.com/books/booksblog/2014/jun/26/lite...

[3] https://github.com/JohannesNE/literature-clock/blob/master/l...

It will also just make up book quotes...

> show book quotes that mention the time 16:36

3. "She glanced at her phone. 16:36. She had just enough time to grab a quick snack before her evening class started. She rummaged through the kitchen and found a granola bar:" (From "Pride and Prejudice" by Jane Austen)

Kmonad for Linux

I'm glad that you seem to get the severity! I'm just hesitant to ascribe malice.

I can only imagine the frustration. Just getting this through peer-review took half a year, but at least there was the academic currency of a publication to motivate me.

The problem is quite subtle, though obvious in retrospect. I've seen a paper from a separate, academic, research group make similar model with the exact same problem.

The problem would, however, have been clear, if the model was compared to simply using the current mean blood pressure (MAP) as a predictor of hypotension, because MAP is the problematic predictor variable. Instead, the model was only compared to short-term changes in MAP (ΔMAP), which is obviously nonsensical and has an AUROC of ~0.55.

Fair question. The model we comment on both suffer from the problem described in the article but also a more severe problem:

The developers sampled obvious cases og hypotension and nonhypotension, and trained the model to distinguish those. And also validated it on data that was similarly dichotomous. In reality the outcome is often between these two scenarios.

But worse, they also introduce a more severe problem where as range of an important predictor is only available in the hypotension outcome.

I recently published a paper, where we explain how an FDA approved prediction model, build into a widely used cardiac monitor was developed with an incredibly biased method.

https://doi.org/10.1097/ALN.0000000000004320

Basically, the training and validation data was engineered so an important range for one of the predictor variables was only present in one of the outcomes, making perfect prediction possible for these cases.

I summarize the paper in this Twitter thread: https://twitter.com/JohsEnevoldsen/status/156164115389992960...

Summary of the paper: https://twitter.com/JohsEnevoldsen/status/156164115389992960...

the imaging tech in both cases is simply a picture of a thin slice of a frozen body. You can navigate the pictures here: https://www.nlm.nih.gov/databases/download/vhp.html

I looked in Additional-Head-Images/cryo/jpeg/fullSize/axial

I think there was also a discovery of some muscle around the knee a few years ago.

compare this to the muscle marked C in this image from the original paper https://ars.els-cdn.com/content/image/1-s2.0-S09409602210020...

I am quite sure you can see it in this and adjecent slices. https://data.lhncbc.nlm.nih.gov/public/Visible-Human/Additio...

Thank you. I think I'll actually use this. A small suggestions: when asking to recall a country name (or flag), only show choices from the already presented countries.

Have you tried switching to Wayland?

I think that is quite sensible, so multiple people do not (unknowingly) spend time writing the same answer.

The latest example of Simpson's Paradox I've seen in the wild was regarding the risk of hospitalization after infection with the B1.1.7 variation of Covid-19 in Denmark. Crude analysis showed that the "old" variation imposed the higher risk. Including age in the model reversed that. B1.1.7 was predominantly spreading among younger people.

Here is a tweet thread explaining this: https://twitter.com/rabaath/status/1335226691304775681?s=20

Really cool. Especially the reference to literature in the differential diagnosis tab. Since the suggested diagnoses are based on case reports, I suspect they are quite biased towards rare diseases and unusual presentations.

One annoying bug: I could not submit 'No' as an answer. It seems to be too short. 'No,' works.

I'm currently working through this book. Highly recommend, even if you have no intention on learning R. The R part is very limited, you will not learn R programming, but if you already know R, it is very useful to end each chapter with a practical demonstration of the theory.

The apple watch ECG corresponds to lead I. Lead II and III is right and left arm respectively to left foot, so they will be possible with a bit of flexibility. The 6 precordial leads are not possible, since they are measuring the voltage between an electrode on the chest and the mean of the 3 extremity electrodes.