TransformEHR: transformer-based encoder-decoder generative model to enhance prediction of disease outcomes using electronic health records
Abstract Deep learning transformer-based models using longitudinal electronic health records (EHRs) have shown a great success in prediction of clinical diseases or outcomes. Pretraining on a large dataset can help such models map the input space better and boost their performance on relevant tasks...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Nature Portfolio
2023-11-01
|
Series: | Nature Communications |
Online Access: | https://doi.org/10.1038/s41467-023-43715-z |
_version_ | 1797414693847957504 |
---|---|
author | Zhichao Yang Avijit Mitra Weisong Liu Dan Berlowitz Hong Yu |
author_facet | Zhichao Yang Avijit Mitra Weisong Liu Dan Berlowitz Hong Yu |
author_sort | Zhichao Yang |
collection | DOAJ |
description | Abstract Deep learning transformer-based models using longitudinal electronic health records (EHRs) have shown a great success in prediction of clinical diseases or outcomes. Pretraining on a large dataset can help such models map the input space better and boost their performance on relevant tasks through finetuning with limited data. In this study, we present TransformEHR, a generative encoder-decoder model with transformer that is pretrained using a new pretraining objective—predicting all diseases and outcomes of a patient at a future visit from previous visits. TransformEHR’s encoder-decoder framework, paired with the novel pretraining objective, helps it achieve the new state-of-the-art performance on multiple clinical prediction tasks. Comparing with the previous model, TransformEHR improves area under the precision–recall curve by 2% (p < 0.001) for pancreatic cancer onset and by 24% (p = 0.007) for intentional self-harm in patients with post-traumatic stress disorder. The high performance in predicting intentional self-harm shows the potential of TransformEHR in building effective clinical intervention systems. TransformEHR is also generalizable and can be easily finetuned for clinical prediction tasks with limited data. |
first_indexed | 2024-03-09T05:37:16Z |
format | Article |
id | doaj.art-e86f3f2a50684c45880702f0abadcb09 |
institution | Directory Open Access Journal |
issn | 2041-1723 |
language | English |
last_indexed | 2024-03-09T05:37:16Z |
publishDate | 2023-11-01 |
publisher | Nature Portfolio |
record_format | Article |
series | Nature Communications |
spelling | doaj.art-e86f3f2a50684c45880702f0abadcb092023-12-03T12:28:03ZengNature PortfolioNature Communications2041-17232023-11-0114111010.1038/s41467-023-43715-zTransformEHR: transformer-based encoder-decoder generative model to enhance prediction of disease outcomes using electronic health recordsZhichao Yang0Avijit Mitra1Weisong Liu2Dan Berlowitz3Hong Yu4College of Information and Computer Science, University of Massachusetts AmherstCollege of Information and Computer Science, University of Massachusetts AmherstSchool of Computer & Information Sciences, University of Massachusetts LowellCenter for Healthcare Organization and Implementation Research, VA Bedford Health Care SystemCollege of Information and Computer Science, University of Massachusetts AmherstAbstract Deep learning transformer-based models using longitudinal electronic health records (EHRs) have shown a great success in prediction of clinical diseases or outcomes. Pretraining on a large dataset can help such models map the input space better and boost their performance on relevant tasks through finetuning with limited data. In this study, we present TransformEHR, a generative encoder-decoder model with transformer that is pretrained using a new pretraining objective—predicting all diseases and outcomes of a patient at a future visit from previous visits. TransformEHR’s encoder-decoder framework, paired with the novel pretraining objective, helps it achieve the new state-of-the-art performance on multiple clinical prediction tasks. Comparing with the previous model, TransformEHR improves area under the precision–recall curve by 2% (p < 0.001) for pancreatic cancer onset and by 24% (p = 0.007) for intentional self-harm in patients with post-traumatic stress disorder. The high performance in predicting intentional self-harm shows the potential of TransformEHR in building effective clinical intervention systems. TransformEHR is also generalizable and can be easily finetuned for clinical prediction tasks with limited data.https://doi.org/10.1038/s41467-023-43715-z |
spellingShingle | Zhichao Yang Avijit Mitra Weisong Liu Dan Berlowitz Hong Yu TransformEHR: transformer-based encoder-decoder generative model to enhance prediction of disease outcomes using electronic health records Nature Communications |
title | TransformEHR: transformer-based encoder-decoder generative model to enhance prediction of disease outcomes using electronic health records |
title_full | TransformEHR: transformer-based encoder-decoder generative model to enhance prediction of disease outcomes using electronic health records |
title_fullStr | TransformEHR: transformer-based encoder-decoder generative model to enhance prediction of disease outcomes using electronic health records |
title_full_unstemmed | TransformEHR: transformer-based encoder-decoder generative model to enhance prediction of disease outcomes using electronic health records |
title_short | TransformEHR: transformer-based encoder-decoder generative model to enhance prediction of disease outcomes using electronic health records |
title_sort | transformehr transformer based encoder decoder generative model to enhance prediction of disease outcomes using electronic health records |
url | https://doi.org/10.1038/s41467-023-43715-z |
work_keys_str_mv | AT zhichaoyang transformehrtransformerbasedencoderdecodergenerativemodeltoenhancepredictionofdiseaseoutcomesusingelectronichealthrecords AT avijitmitra transformehrtransformerbasedencoderdecodergenerativemodeltoenhancepredictionofdiseaseoutcomesusingelectronichealthrecords AT weisongliu transformehrtransformerbasedencoderdecodergenerativemodeltoenhancepredictionofdiseaseoutcomesusingelectronichealthrecords AT danberlowitz transformehrtransformerbasedencoderdecodergenerativemodeltoenhancepredictionofdiseaseoutcomesusingelectronichealthrecords AT hongyu transformehrtransformerbasedencoderdecodergenerativemodeltoenhancepredictionofdiseaseoutcomesusingelectronichealthrecords |