Closed alxndrkalinin closed 7 years ago
Good question. We haven't decided anything yet. Defining the most common neural network abbreviations like those in the issue title in the introduction makes sense. Others can be defined at first use.
@cgreene let us know if you have thoughts.
NDC -> National Drug Codes (came up in #339)
ICD -> International Classification of Diseases
I figure we'll need to go through and define acronyms the first time they appear. If you encounter some, feel free to note them here.
NLP: natural language processing (#342) - Plan to introduce it in Categorize/EHR, I can see it's also used in Study/Metagenomics. GWAS: genome-wide association study (#342) - Also used later on.
RF -> random forest
May decide to always write it out in full, but we should be consistent throughout.
What I've spotted that appear to occur more than once:
EHR: electronic health record RNN: recurrent NN TF: transcription factor MRI / fMRI: Magnetic resonance imaging DBM: deep boltzman machine DBN: deep belief network MTL: multi-task learning
AUC: area under the curve ROC: receiver operating characteristic
Needed for #438 among others. This is also called auROC
in parts of the review.
I'm going through these in #498 and adding 👍 once I cover them.
Instead of abbreviating random forest as RF, which I don't care for, I'm using square brackets to write it out in a quote. This will show up as a warning during builds, but we can ignore it.
Couldn't find any guidelines on abbreviations - are we going to introduce common ones like CNN in the beginning of the manuscript and use further on? Is there journal policy on this?