Deep Learning Model for Predicting Intradialytic Hypotension Without Privacy Infringement: A Retrospective Two-Center Study
ObjectivePreviously developed Intradialytic hypotension (IDH) prediction models utilize clinical variables with potential privacy protection issues. We developed an IDH prediction model using minimal variables, without the risk of privacy infringement.MethodsUnidentifiable data from 63,640 hemodialy...
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2022-07-01
|
| Series: | Frontiers in Medicine |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fmed.2022.878858/full |
| _version_ | 1818466493807984640 |
|---|---|
| author | Hyung Woo Kim Seok-Jae Heo Minseok Kim Jakyung Lee Keun Hyung Park Gongmyung Lee Song In Baeg Young Eun Kwon Hye Min Choi Dong-Jin Oh Chung-Mo Nam Chung-Mo Nam Beom Seok Kim |
| author_facet | Hyung Woo Kim Seok-Jae Heo Minseok Kim Jakyung Lee Keun Hyung Park Gongmyung Lee Song In Baeg Young Eun Kwon Hye Min Choi Dong-Jin Oh Chung-Mo Nam Chung-Mo Nam Beom Seok Kim |
| author_sort | Hyung Woo Kim |
| collection | DOAJ |
| description | ObjectivePreviously developed Intradialytic hypotension (IDH) prediction models utilize clinical variables with potential privacy protection issues. We developed an IDH prediction model using minimal variables, without the risk of privacy infringement.MethodsUnidentifiable data from 63,640 hemodialysis sessions (26,746 of 79 patients for internal validation, 36,894 of 255 patients for external validation) from two Korean hospital hemodialysis databases were finally analyzed, using three IDH definitions: (1) systolic blood pressure (SBP) nadir <90 mmHg (Nadir90); (2) SBP decrease ≥20 mmHg from baseline (Fall20); and (3) SBP decrease ≥20 mmHg and/or mean arterial pressure decrease ≥10 mmHg (Fall20/MAP10). The developed models use 30 min information to predict an IDH event in the following 10 min window. Area under the receiver operating characteristic curves (AUROCs) and precision-recall curves were used to compare machine learning and deep learning models by logistic regression, XGBoost, and convolutional neural networks.ResultsAmong 344,714 segments, 9,154 (2.7%), 134,988 (39.2%), and 149,674 (43.4%) IDH events occurred according to three different IDH definitions (Nadir90, Fall20, and Fall20/MAP10, respectively). Compared with models including logistic regression, random forest, and XGBoost, the deep learning model achieved the best performance in predicting IDH (AUROCs: Nadir90, 0.905; Fall20, 0.864; Fall20/MAP10, 0.863) only using measurements from hemodialysis machine during dialysis session.ConclusionsThe deep learning model performed well only using monitoring measurement of hemodialysis machine in predicting IDH without any personal information that could risk privacy infringement. |
| first_indexed | 2024-04-13T20:59:10Z |
| format | Article |
| id | doaj.art-8faa5780f14a449b84d3f4e7b0d612a3 |
| institution | Directory Open Access Journal |
| issn | 2296-858X |
| language | English |
| last_indexed | 2024-04-13T20:59:10Z |
| publishDate | 2022-07-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Medicine |
| spelling | doaj.art-8faa5780f14a449b84d3f4e7b0d612a32022-12-22T02:30:13ZengFrontiers Media S.A.Frontiers in Medicine2296-858X2022-07-01910.3389/fmed.2022.878858878858Deep Learning Model for Predicting Intradialytic Hypotension Without Privacy Infringement: A Retrospective Two-Center StudyHyung Woo Kim0Seok-Jae Heo1Minseok Kim2Jakyung Lee3Keun Hyung Park4Gongmyung Lee5Song In Baeg6Young Eun Kwon7Hye Min Choi8Dong-Jin Oh9Chung-Mo Nam10Chung-Mo Nam11Beom Seok Kim12Department of Internal Medicine, Yonsei University College of Medicine, Seoul, South KoreaDepartment of Biostatistics and Computing, Yonsei University Graduate School, Seoul, South KoreaDepartment of Biostatistics and Computing, Yonsei University Graduate School, Seoul, South KoreaDepartment of Biostatistics and Computing, Yonsei University Graduate School, Seoul, South KoreaDepartment of Internal Medicine, Yonsei University College of Medicine, Seoul, South KoreaDepartment of Internal Medicine, Yonsei University College of Medicine, Seoul, South KoreaDepartment of Internal Medicine, Myongji Hospital, Hanyang University College of Medicine, Goyang, South KoreaDepartment of Internal Medicine, Myongji Hospital, Hanyang University College of Medicine, Goyang, South KoreaDepartment of Internal Medicine, Myongji Hospital, Hanyang University College of Medicine, Goyang, South KoreaDepartment of Internal Medicine, Myongji Hospital, Hanyang University College of Medicine, Goyang, South KoreaDepartment of Biostatistics and Computing, Yonsei University Graduate School, Seoul, South KoreaDivision of Biostatistics, Department of Biomedical Systems Informatics, Yonsei University College of Medicine, Seoul, South KoreaDepartment of Internal Medicine, Yonsei University College of Medicine, Seoul, South KoreaObjectivePreviously developed Intradialytic hypotension (IDH) prediction models utilize clinical variables with potential privacy protection issues. We developed an IDH prediction model using minimal variables, without the risk of privacy infringement.MethodsUnidentifiable data from 63,640 hemodialysis sessions (26,746 of 79 patients for internal validation, 36,894 of 255 patients for external validation) from two Korean hospital hemodialysis databases were finally analyzed, using three IDH definitions: (1) systolic blood pressure (SBP) nadir <90 mmHg (Nadir90); (2) SBP decrease ≥20 mmHg from baseline (Fall20); and (3) SBP decrease ≥20 mmHg and/or mean arterial pressure decrease ≥10 mmHg (Fall20/MAP10). The developed models use 30 min information to predict an IDH event in the following 10 min window. Area under the receiver operating characteristic curves (AUROCs) and precision-recall curves were used to compare machine learning and deep learning models by logistic regression, XGBoost, and convolutional neural networks.ResultsAmong 344,714 segments, 9,154 (2.7%), 134,988 (39.2%), and 149,674 (43.4%) IDH events occurred according to three different IDH definitions (Nadir90, Fall20, and Fall20/MAP10, respectively). Compared with models including logistic regression, random forest, and XGBoost, the deep learning model achieved the best performance in predicting IDH (AUROCs: Nadir90, 0.905; Fall20, 0.864; Fall20/MAP10, 0.863) only using measurements from hemodialysis machine during dialysis session.ConclusionsThe deep learning model performed well only using monitoring measurement of hemodialysis machine in predicting IDH without any personal information that could risk privacy infringement.https://www.frontiersin.org/articles/10.3389/fmed.2022.878858/fulldeep learningintradialytic hypotensionmachine learningprivacy protectionhemodialysis |
| spellingShingle | Hyung Woo Kim Seok-Jae Heo Minseok Kim Jakyung Lee Keun Hyung Park Gongmyung Lee Song In Baeg Young Eun Kwon Hye Min Choi Dong-Jin Oh Chung-Mo Nam Chung-Mo Nam Beom Seok Kim Deep Learning Model for Predicting Intradialytic Hypotension Without Privacy Infringement: A Retrospective Two-Center Study Frontiers in Medicine deep learning intradialytic hypotension machine learning privacy protection hemodialysis |
| title | Deep Learning Model for Predicting Intradialytic Hypotension Without Privacy Infringement: A Retrospective Two-Center Study |
| title_full | Deep Learning Model for Predicting Intradialytic Hypotension Without Privacy Infringement: A Retrospective Two-Center Study |
| title_fullStr | Deep Learning Model for Predicting Intradialytic Hypotension Without Privacy Infringement: A Retrospective Two-Center Study |
| title_full_unstemmed | Deep Learning Model for Predicting Intradialytic Hypotension Without Privacy Infringement: A Retrospective Two-Center Study |
| title_short | Deep Learning Model for Predicting Intradialytic Hypotension Without Privacy Infringement: A Retrospective Two-Center Study |
| title_sort | deep learning model for predicting intradialytic hypotension without privacy infringement a retrospective two center study |
| topic | deep learning intradialytic hypotension machine learning privacy protection hemodialysis |
| url | https://www.frontiersin.org/articles/10.3389/fmed.2022.878858/full |
| work_keys_str_mv | AT hyungwookim deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT seokjaeheo deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT minseokkim deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT jakyunglee deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT keunhyungpark deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT gongmyunglee deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT songinbaeg deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT youngeunkwon deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT hyeminchoi deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT dongjinoh deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT chungmonam deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT chungmonam deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy AT beomseokkim deeplearningmodelforpredictingintradialytichypotensionwithoutprivacyinfringementaretrospectivetwocenterstudy |