We develop the first large-scale algorithm to automatically predict all 4577 standard veterinary diagnosis codes from free text.
We trained our algorithm on a curated dataset of over 100K expert labeled veterinary clinical notes and over one million unlabeled notes.
In order to better understand how DeepVet predicts diagnosis codes from clinical notes, we implement a simple saliency-based interpretation method
$PET_NAME$, a 2 year old male neutered boxer , presented to CSU dermatology service for a recheck of allergic dermatitis . It was reported that $PET_NAME$ 's pruritus is relatively seasonal , worse in the spring and summer , but does not completely go away in the winter . $PET_NAME$'s main problem areas were his paws and his muzzle .
DeepVet also learns to abstain in examples when it is uncertain and defers them to human experts, fostering human-machine collaboration.
In order to not mislead further clinical research, having the ability to abstain from making very erroneous predictions and ensuring highly precise tagging is an important feature.
Large-scale Generative Modeling to Improve Automated Veterinary Disease Coding.
Yuhui Zhang, Allen Nie, and James Zou.
NeurIPS Machine Learning for Health Workshop (2018).
DeepTag: inferring diagnoses from veterinary clinical notes.
Allen Nie, Ashley Zehnder, Rodney L. Page, Yuhui Zhang, Arturo Lopez Pineda, Manuel A. Rivas, Carlos D. Bustamante, and James Zou.
npj Digital Medicine 1, no. 1 (2018): 60.