Clinical Relevance Score for Guided Trauma Injury Pattern Discovery with Weakly Supervised $β$-VAE

Qixuan Jin, Jacobien HF Oosterhoff, Yepeng Huang, Marzyeh Ghassemi, Gabriel A Brat
Proceedings of the Conference on Health, Inference, and Learning, PMLR 209:314-339, 2023.

Abstract

Given the complexity of trauma presentations, particularly in those involving multiple areas of the body, overlooked injuries are common during the initial assessment by a clinician. We are motivated to develop an automated trauma pattern discovery framework for comprehensive identification of injury patterns which may eventually support diagnostic decision-making. We analyze 1,162,399 patients from the Trauma Quality Improvement Program with a disentangled variational autoencoder, weakly supervised by a latent-space classifier of auxiliary features. We also develop a novel scoring metric that serves as a proxy for clinical intuition in extracting clusters with clinically meaningful injury patterns. We validate the extracted clusters with clinical experts, and explore the patient characteristics of selected groupings. Our metric is able to perform model selection and effectively filter clusters for clinically-validated relevance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v209-jin23a, title = {Clinical Relevance Score for Guided Trauma Injury Pattern Discovery with Weakly Supervised $β$-VAE}, author = {Jin, Qixuan and Oosterhoff, Jacobien HF and Huang, Yepeng and Ghassemi, Marzyeh and Brat, Gabriel A}, booktitle = {Proceedings of the Conference on Health, Inference, and Learning}, pages = {314--339}, year = {2023}, editor = {Mortazavi, Bobak J. and Sarker, Tasmie and Beam, Andrew and Ho, Joyce C.}, volume = {209}, series = {Proceedings of Machine Learning Research}, month = {22 Jun--24 Jun}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v209/jin23a/jin23a.pdf}, url = {https://proceedings.mlr.press/v209/jin23a.html}, abstract = {Given the complexity of trauma presentations, particularly in those involving multiple areas of the body, overlooked injuries are common during the initial assessment by a clinician. We are motivated to develop an automated trauma pattern discovery framework for comprehensive identification of injury patterns which may eventually support diagnostic decision-making. We analyze 1,162,399 patients from the Trauma Quality Improvement Program with a disentangled variational autoencoder, weakly supervised by a latent-space classifier of auxiliary features. We also develop a novel scoring metric that serves as a proxy for clinical intuition in extracting clusters with clinically meaningful injury patterns. We validate the extracted clusters with clinical experts, and explore the patient characteristics of selected groupings. Our metric is able to perform model selection and effectively filter clusters for clinically-validated relevance. } }
Endnote
%0 Conference Paper %T Clinical Relevance Score for Guided Trauma Injury Pattern Discovery with Weakly Supervised $β$-VAE %A Qixuan Jin %A Jacobien HF Oosterhoff %A Yepeng Huang %A Marzyeh Ghassemi %A Gabriel A Brat %B Proceedings of the Conference on Health, Inference, and Learning %C Proceedings of Machine Learning Research %D 2023 %E Bobak J. Mortazavi %E Tasmie Sarker %E Andrew Beam %E Joyce C. Ho %F pmlr-v209-jin23a %I PMLR %P 314--339 %U https://proceedings.mlr.press/v209/jin23a.html %V 209 %X Given the complexity of trauma presentations, particularly in those involving multiple areas of the body, overlooked injuries are common during the initial assessment by a clinician. We are motivated to develop an automated trauma pattern discovery framework for comprehensive identification of injury patterns which may eventually support diagnostic decision-making. We analyze 1,162,399 patients from the Trauma Quality Improvement Program with a disentangled variational autoencoder, weakly supervised by a latent-space classifier of auxiliary features. We also develop a novel scoring metric that serves as a proxy for clinical intuition in extracting clusters with clinically meaningful injury patterns. We validate the extracted clusters with clinical experts, and explore the patient characteristics of selected groupings. Our metric is able to perform model selection and effectively filter clusters for clinically-validated relevance.
APA
Jin, Q., Oosterhoff, J.H., Huang, Y., Ghassemi, M. & Brat, G.A.. (2023). Clinical Relevance Score for Guided Trauma Injury Pattern Discovery with Weakly Supervised $β$-VAE. Proceedings of the Conference on Health, Inference, and Learning, in Proceedings of Machine Learning Research 209:314-339 Available from https://proceedings.mlr.press/v209/jin23a.html.

Related Material