Robust Fusion of Time Series and Image Data for Improved Multimodal Clinical Prediction

With the increasing availability of diverse data types, particularly images and time series data from medical experiments, there is a growing demand for techniques designed to combine various modalities of data effectively. Our motivation comes from the important areas of predicting mortality and ph...

Full description

Saved in:
Bibliographic Details
Main Authors: Ali Rasekh, Reza Heidari, Amir Hosein Haji Mohammad Rezaie, Parsa Sharifi Sedeh, Zahra Ahmadi, Prasenjit Mitra, Wolfgang Nejdl
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10752509/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850267878168723456
author Ali Rasekh
Reza Heidari
Amir Hosein Haji Mohammad Rezaie
Parsa Sharifi Sedeh
Zahra Ahmadi
Prasenjit Mitra
Wolfgang Nejdl
author_facet Ali Rasekh
Reza Heidari
Amir Hosein Haji Mohammad Rezaie
Parsa Sharifi Sedeh
Zahra Ahmadi
Prasenjit Mitra
Wolfgang Nejdl
author_sort Ali Rasekh
collection DOAJ
description With the increasing availability of diverse data types, particularly images and time series data from medical experiments, there is a growing demand for techniques designed to combine various modalities of data effectively. Our motivation comes from the important areas of predicting mortality and phenotyping where using different modalities of data could significantly improve our ability to predict. To tackle this challenge, we introduce a new method that uses two separate encoders, one for each type of data, allowing the model to understand complex patterns in both visual and time-based information. Apart from the technical challenges, our goal is to make the predictive model more robust in noisy conditions and perform better than current methods. We also deal with imbalanced datasets and use an uncertainty loss function, yielding improved results while simultaneously providing a principled means of modeling uncertainty. Additionally, we include attention mechanisms to fuse different modalities, allowing the model to focus on what&#x2019;s important for each task. We tested our approach using the comprehensive multimodal MIMIC dataset, combining MIMIC-IV and MIMIC-CXR datasets. Our experiments show that our method is effective in improving multimodal deep learning for clinical applications. The code for this work is publicly available at: <uri>https://github.com/AliRasekh/TSImageFusion</uri>
format Article
id doaj-art-ff7d69c3aead48a3aad623b80c2bca03
institution OA Journals
issn 2169-3536
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-ff7d69c3aead48a3aad623b80c2bca032025-08-20T01:53:36ZengIEEEIEEE Access2169-35362024-01-011217410717412110.1109/ACCESS.2024.349766810752509Robust Fusion of Time Series and Image Data for Improved Multimodal Clinical PredictionAli Rasekh0https://orcid.org/0009-0005-6808-6730Reza Heidari1https://orcid.org/0009-0007-6260-6592Amir Hosein Haji Mohammad Rezaie2https://orcid.org/0009-0008-5072-971XParsa Sharifi Sedeh3https://orcid.org/0009-0000-3621-1887Zahra Ahmadi4Prasenjit Mitra5Wolfgang Nejdl6L3S Research Center, Leibniz University Hannover, Hannover, GermanyL3S Research Center, Leibniz University Hannover, Hannover, GermanyL3S Research Center, Leibniz University Hannover, Hannover, GermanyL3S Research Center, Leibniz University Hannover, Hannover, GermanyL3S Research Center, Leibniz University Hannover, Hannover, GermanyL3S Research Center, Leibniz University Hannover, Hannover, GermanyL3S Research Center, Leibniz University Hannover, Hannover, GermanyWith the increasing availability of diverse data types, particularly images and time series data from medical experiments, there is a growing demand for techniques designed to combine various modalities of data effectively. Our motivation comes from the important areas of predicting mortality and phenotyping where using different modalities of data could significantly improve our ability to predict. To tackle this challenge, we introduce a new method that uses two separate encoders, one for each type of data, allowing the model to understand complex patterns in both visual and time-based information. Apart from the technical challenges, our goal is to make the predictive model more robust in noisy conditions and perform better than current methods. We also deal with imbalanced datasets and use an uncertainty loss function, yielding improved results while simultaneously providing a principled means of modeling uncertainty. Additionally, we include attention mechanisms to fuse different modalities, allowing the model to focus on what&#x2019;s important for each task. We tested our approach using the comprehensive multimodal MIMIC dataset, combining MIMIC-IV and MIMIC-CXR datasets. Our experiments show that our method is effective in improving multimodal deep learning for clinical applications. The code for this work is publicly available at: <uri>https://github.com/AliRasekh/TSImageFusion</uri>https://ieeexplore.ieee.org/document/10752509/Multimodal learningtime seriesattention mechanismrobustnessphenotyping
spellingShingle Ali Rasekh
Reza Heidari
Amir Hosein Haji Mohammad Rezaie
Parsa Sharifi Sedeh
Zahra Ahmadi
Prasenjit Mitra
Wolfgang Nejdl
Robust Fusion of Time Series and Image Data for Improved Multimodal Clinical Prediction
IEEE Access
Multimodal learning
time series
attention mechanism
robustness
phenotyping
title Robust Fusion of Time Series and Image Data for Improved Multimodal Clinical Prediction
title_full Robust Fusion of Time Series and Image Data for Improved Multimodal Clinical Prediction
title_fullStr Robust Fusion of Time Series and Image Data for Improved Multimodal Clinical Prediction
title_full_unstemmed Robust Fusion of Time Series and Image Data for Improved Multimodal Clinical Prediction
title_short Robust Fusion of Time Series and Image Data for Improved Multimodal Clinical Prediction
title_sort robust fusion of time series and image data for improved multimodal clinical prediction
topic Multimodal learning
time series
attention mechanism
robustness
phenotyping
url https://ieeexplore.ieee.org/document/10752509/
work_keys_str_mv AT alirasekh robustfusionoftimeseriesandimagedataforimprovedmultimodalclinicalprediction
AT rezaheidari robustfusionoftimeseriesandimagedataforimprovedmultimodalclinicalprediction
AT amirhoseinhajimohammadrezaie robustfusionoftimeseriesandimagedataforimprovedmultimodalclinicalprediction
AT parsasharifisedeh robustfusionoftimeseriesandimagedataforimprovedmultimodalclinicalprediction
AT zahraahmadi robustfusionoftimeseriesandimagedataforimprovedmultimodalclinicalprediction
AT prasenjitmitra robustfusionoftimeseriesandimagedataforimprovedmultimodalclinicalprediction
AT wolfgangnejdl robustfusionoftimeseriesandimagedataforimprovedmultimodalclinicalprediction