Simultaneous forecasting of vital sign trajectories in the ICU

Abstract Individual health trajectory forecasting is a major opportunity for computational methods to integrate with precision healthcare. Recently developed generative AI models have demonstrated promising results in capturing short and long range dependencies in time series data. While these model...

Full description

Saved in:
Bibliographic Details
Main Authors: Rosemary He, Jeffrey N. Chiang
Format: Article
Language:English
Published: Nature Portfolio 2025-04-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-99719-w
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849314719894077440
author Rosemary He
Jeffrey N. Chiang
author_facet Rosemary He
Jeffrey N. Chiang
author_sort Rosemary He
collection DOAJ
description Abstract Individual health trajectory forecasting is a major opportunity for computational methods to integrate with precision healthcare. Recently developed generative AI models have demonstrated promising results in capturing short and long range dependencies in time series data. While these models have also been applied in healthcare, most state-of-the-art are local models, i.e. one model per feature, which is unrealistic in a clinical setting where multiple measures are taken at once. In this work, we extend the framework temporal fusion transformer (TFT), a multi-horizon time series prediction tool, and propose TFT-multi, a global model that can predict multiple vital trajectories simultaneously. We apply TFT-multi to forecast 5 vital signs recorded in the intensive care unit: blood pressure, pulse, SpO2, temperature and respiratory rate. We hypothesize that by jointly predicting these measures, which are often correlated with one another, we can make more accurate predictions, especially in variables with large missingness. We validate our model on the public MIMIC dataset and an independent institutional dataset, and demonstrate our model’s competitive performance and computational efficiency compared to state-of-the-art prediction tools. Furthermore, we perform a study case analysis by applying our pipeline to forecast blood pressure changes in response to actual and hypothetical pressor administration.
format Article
id doaj-art-a529bc45808d4d42845ca18d76740416
institution Kabale University
issn 2045-2322
language English
publishDate 2025-04-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-a529bc45808d4d42845ca18d767404162025-08-20T03:52:23ZengNature PortfolioScientific Reports2045-23222025-04-0115111010.1038/s41598-025-99719-wSimultaneous forecasting of vital sign trajectories in the ICURosemary He0Jeffrey N. Chiang1Department of Computer Science, University of California Los AngelesDepartment of Computational Medicine, University of California Los AngelesAbstract Individual health trajectory forecasting is a major opportunity for computational methods to integrate with precision healthcare. Recently developed generative AI models have demonstrated promising results in capturing short and long range dependencies in time series data. While these models have also been applied in healthcare, most state-of-the-art are local models, i.e. one model per feature, which is unrealistic in a clinical setting where multiple measures are taken at once. In this work, we extend the framework temporal fusion transformer (TFT), a multi-horizon time series prediction tool, and propose TFT-multi, a global model that can predict multiple vital trajectories simultaneously. We apply TFT-multi to forecast 5 vital signs recorded in the intensive care unit: blood pressure, pulse, SpO2, temperature and respiratory rate. We hypothesize that by jointly predicting these measures, which are often correlated with one another, we can make more accurate predictions, especially in variables with large missingness. We validate our model on the public MIMIC dataset and an independent institutional dataset, and demonstrate our model’s competitive performance and computational efficiency compared to state-of-the-art prediction tools. Furthermore, we perform a study case analysis by applying our pipeline to forecast blood pressure changes in response to actual and hypothetical pressor administration.https://doi.org/10.1038/s41598-025-99719-wGlobal trajectory predictionsGenerative AITime seriesClinical support
spellingShingle Rosemary He
Jeffrey N. Chiang
Simultaneous forecasting of vital sign trajectories in the ICU
Scientific Reports
Global trajectory predictions
Generative AI
Time series
Clinical support
title Simultaneous forecasting of vital sign trajectories in the ICU
title_full Simultaneous forecasting of vital sign trajectories in the ICU
title_fullStr Simultaneous forecasting of vital sign trajectories in the ICU
title_full_unstemmed Simultaneous forecasting of vital sign trajectories in the ICU
title_short Simultaneous forecasting of vital sign trajectories in the ICU
title_sort simultaneous forecasting of vital sign trajectories in the icu
topic Global trajectory predictions
Generative AI
Time series
Clinical support
url https://doi.org/10.1038/s41598-025-99719-w
work_keys_str_mv AT rosemaryhe simultaneousforecastingofvitalsigntrajectoriesintheicu
AT jeffreynchiang simultaneousforecastingofvitalsigntrajectoriesintheicu