Hybrid machine learning for real-time prediction of edema trajectory in large middle cerebral artery stroke

Abstract In treating malignant cerebral edema after a large middle cerebral artery stroke, clinicians need quantitative tools for real-time risk assessment. Existing predictive models typically estimate risk at one, early time point, failing to account for dynamic variables. To address this, we deve...

Full description

Saved in:
Bibliographic Details
Main Authors: Ethan Phillips, Odhran O’Donoghue, Yumeng Zhang, Panos Tsimpos, Leigh Ann Mallinger, Stefanos Chatzidakis, Jack Pohlmann, Yili Du, Ivy Kim, Jonathan Song, Benjamin Brush, Stelios Smirnakis, Charlene J. Ong, Agni Orfanoudaki
Format: Article
Language:English
Published: Nature Portfolio 2025-05-01
Series:npj Digital Medicine
Online Access:https://doi.org/10.1038/s41746-025-01687-y
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850154503002652672
author Ethan Phillips
Odhran O’Donoghue
Yumeng Zhang
Panos Tsimpos
Leigh Ann Mallinger
Stefanos Chatzidakis
Jack Pohlmann
Yili Du
Ivy Kim
Jonathan Song
Benjamin Brush
Stelios Smirnakis
Charlene J. Ong
Agni Orfanoudaki
author_facet Ethan Phillips
Odhran O’Donoghue
Yumeng Zhang
Panos Tsimpos
Leigh Ann Mallinger
Stefanos Chatzidakis
Jack Pohlmann
Yili Du
Ivy Kim
Jonathan Song
Benjamin Brush
Stelios Smirnakis
Charlene J. Ong
Agni Orfanoudaki
author_sort Ethan Phillips
collection DOAJ
description Abstract In treating malignant cerebral edema after a large middle cerebral artery stroke, clinicians need quantitative tools for real-time risk assessment. Existing predictive models typically estimate risk at one, early time point, failing to account for dynamic variables. To address this, we developed Hybrid Ensemble Learning Models for Edema Trajectory (HELMET) to predict midline shift severity, an established indicator of malignant edema, over 8-h and 24-h windows. The HELMET models were trained on retrospective data from 623 patients and validated on 63 patients from a different hospital system, achieving mean areas under the receiver operating characteristic curve of 96.6% and 92.5%, respectively. By integrating transformer-based large language models with supervised ensemble learning, HELMET demonstrates the value of combining clinician expertise with multimodal health records in assessing patient risk. Our approach provides a framework for accurate, real-time estimation of dynamic clinical targets using human-curated and algorithm-derived inputs.
format Article
id doaj-art-e03065317dfb40e19a923851c1ab26bb
institution OA Journals
issn 2398-6352
language English
publishDate 2025-05-01
publisher Nature Portfolio
record_format Article
series npj Digital Medicine
spelling doaj-art-e03065317dfb40e19a923851c1ab26bb2025-08-20T02:25:17ZengNature Portfolionpj Digital Medicine2398-63522025-05-018111410.1038/s41746-025-01687-yHybrid machine learning for real-time prediction of edema trajectory in large middle cerebral artery strokeEthan Phillips0Odhran O’Donoghue1Yumeng Zhang2Panos Tsimpos3Leigh Ann Mallinger4Stefanos Chatzidakis5Jack Pohlmann6Yili Du7Ivy Kim8Jonathan Song9Benjamin Brush10Stelios Smirnakis11Charlene J. Ong12Agni Orfanoudaki13University of OxfordUniversity of OxfordNorth Carolina State UniversityMassachusetts Institute of TechnologyBoston Medical Center, Department of NeurologyBrigham & Women’s Hospital, Department of NeurologyBoston Medical Center, Department of NeurologyBoston University School of Public HealthBoston Medical Center, Department of NeurologyBoston University Chobanian & Avedisian School of MedicineNYU Langone HospitalBrigham & Women’s Hospital, Department of NeurologyBoston Medical Center, Department of NeurologyUniversity of OxfordAbstract In treating malignant cerebral edema after a large middle cerebral artery stroke, clinicians need quantitative tools for real-time risk assessment. Existing predictive models typically estimate risk at one, early time point, failing to account for dynamic variables. To address this, we developed Hybrid Ensemble Learning Models for Edema Trajectory (HELMET) to predict midline shift severity, an established indicator of malignant edema, over 8-h and 24-h windows. The HELMET models were trained on retrospective data from 623 patients and validated on 63 patients from a different hospital system, achieving mean areas under the receiver operating characteristic curve of 96.6% and 92.5%, respectively. By integrating transformer-based large language models with supervised ensemble learning, HELMET demonstrates the value of combining clinician expertise with multimodal health records in assessing patient risk. Our approach provides a framework for accurate, real-time estimation of dynamic clinical targets using human-curated and algorithm-derived inputs.https://doi.org/10.1038/s41746-025-01687-y
spellingShingle Ethan Phillips
Odhran O’Donoghue
Yumeng Zhang
Panos Tsimpos
Leigh Ann Mallinger
Stefanos Chatzidakis
Jack Pohlmann
Yili Du
Ivy Kim
Jonathan Song
Benjamin Brush
Stelios Smirnakis
Charlene J. Ong
Agni Orfanoudaki
Hybrid machine learning for real-time prediction of edema trajectory in large middle cerebral artery stroke
npj Digital Medicine
title Hybrid machine learning for real-time prediction of edema trajectory in large middle cerebral artery stroke
title_full Hybrid machine learning for real-time prediction of edema trajectory in large middle cerebral artery stroke
title_fullStr Hybrid machine learning for real-time prediction of edema trajectory in large middle cerebral artery stroke
title_full_unstemmed Hybrid machine learning for real-time prediction of edema trajectory in large middle cerebral artery stroke
title_short Hybrid machine learning for real-time prediction of edema trajectory in large middle cerebral artery stroke
title_sort hybrid machine learning for real time prediction of edema trajectory in large middle cerebral artery stroke
url https://doi.org/10.1038/s41746-025-01687-y
work_keys_str_mv AT ethanphillips hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT odhranodonoghue hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT yumengzhang hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT panostsimpos hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT leighannmallinger hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT stefanoschatzidakis hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT jackpohlmann hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT yilidu hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT ivykim hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT jonathansong hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT benjaminbrush hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT steliossmirnakis hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT charlenejong hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke
AT agniorfanoudaki hybridmachinelearningforrealtimepredictionofedematrajectoryinlargemiddlecerebralarterystroke