Saved in:
Bibliographic Details
Main Authors: Xue, Yuan, Du, Nan, Mottram, Anne, Seneviratne, Martin, Dai, Andrew M.
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2407.19359
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909270711205888
author Xue, Yuan
Du, Nan
Mottram, Anne
Seneviratne, Martin
Dai, Andrew M.
author_facet Xue, Yuan
Du, Nan
Mottram, Anne
Seneviratne, Martin
Dai, Andrew M.
contents We propose to meta-learn an a self-supervised patient trajectory forecast learning rule by meta-training on a meta-objective that directly optimizes the utility of the patient representation over the subsequent clinical outcome prediction. This meta-objective directly targets the usefulness of a representation generated from unlabeled clinical measurement forecast for later supervised tasks. The meta-learned can then be directly used in target risk prediction, and the limited available samples can be used for further fine-tuning the model performance. The effectiveness of our approach is tested on a real open source patient EHR dataset MIMIC-III. We are able to demonstrate that our attention-based patient state representation approach can achieve much better performance for predicting target risk with low resources comparing with both direct supervised learning and pretraining with all-observation trajectory forecast.
format Preprint
id arxiv_https___arxiv_org_abs_2407_19359
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Learning to Select the Best Forecasting Tasks for Clinical Outcome Prediction
Xue, Yuan
Du, Nan
Mottram, Anne
Seneviratne, Martin
Dai, Andrew M.
Machine Learning
Artificial Intelligence
We propose to meta-learn an a self-supervised patient trajectory forecast learning rule by meta-training on a meta-objective that directly optimizes the utility of the patient representation over the subsequent clinical outcome prediction. This meta-objective directly targets the usefulness of a representation generated from unlabeled clinical measurement forecast for later supervised tasks. The meta-learned can then be directly used in target risk prediction, and the limited available samples can be used for further fine-tuning the model performance. The effectiveness of our approach is tested on a real open source patient EHR dataset MIMIC-III. We are able to demonstrate that our attention-based patient state representation approach can achieve much better performance for predicting target risk with low resources comparing with both direct supervised learning and pretraining with all-observation trajectory forecast.
title Learning to Select the Best Forecasting Tasks for Clinical Outcome Prediction
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2407.19359