ICU Survival Prediction Incorporating Test-Time Augmentation to Improve the Accuracy of Ensemble-Based Models
This work presents a novel method for applying test-time augmentation (TTA) to tabular data. We used TTA along with an ensemble of 42 models to achieve higher performance on the MIT Global Open Source Severity of Illness Score dataset consisting of 131,051 ICU visits and outcomes. This method achiev...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2021-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9462159/ |
| id |
doaj-ebc84f1edfb04aaab40429ff8463a941 |
|---|---|
| record_format |
Article |
| spelling |
doaj-ebc84f1edfb04aaab40429ff8463a9412021-06-30T23:00:09ZengIEEEIEEE Access2169-35362021-01-019915849159210.1109/ACCESS.2021.30916229462159ICU Survival Prediction Incorporating Test-Time Augmentation to Improve the Accuracy of Ensemble-Based ModelsSeffi Cohen0https://orcid.org/0000-0002-1135-0079Noa Dagan1Nurit Cohen-Inger2Dan Ofer3https://orcid.org/0000-0001-5136-8014Lior Rokach4Department of Software and Information Systems Engineering, Ben-Gurion University, Beer-Sheva, IsraelDepartment of Software and Information Systems Engineering, Ben-Gurion University, Beer-Sheva, IsraelBeyondMinds, Tel-Aviv, IsraelMedtronic, Tel-Aviv, IsraelDepartment of Software and Information Systems Engineering, Ben-Gurion University, Beer-Sheva, IsraelThis work presents a novel method for applying test-time augmentation (TTA) to tabular data. We used TTA along with an ensemble of 42 models to achieve higher performance on the MIT Global Open Source Severity of Illness Score dataset consisting of 131,051 ICU visits and outcomes. This method achieved an AUC of 0.915 on the private test set (19,669 admissions) and won first place at Stanford University’s WiDS Datathon 2020 challenge on Kaggle, while the Acute Physiology and Chronic Health Evaluation (APACHE) IV model (commonly used for ICU survival prediction in the literature) achieved an AUC of 0.868. In addition to increasing the AUC score, our method also reduces “unfair” bias.https://ieeexplore.ieee.org/document/9462159/Ensemble methodsmachine learningsupervised classificationhealthcare |
| collection |
DOAJ |
| language |
English |
| format |
Article |
| sources |
DOAJ |
| author |
Seffi Cohen Noa Dagan Nurit Cohen-Inger Dan Ofer Lior Rokach |
| spellingShingle |
Seffi Cohen Noa Dagan Nurit Cohen-Inger Dan Ofer Lior Rokach ICU Survival Prediction Incorporating Test-Time Augmentation to Improve the Accuracy of Ensemble-Based Models IEEE Access Ensemble methods machine learning supervised classification healthcare |
| author_facet |
Seffi Cohen Noa Dagan Nurit Cohen-Inger Dan Ofer Lior Rokach |
| author_sort |
Seffi Cohen |
| title |
ICU Survival Prediction Incorporating Test-Time Augmentation to Improve the Accuracy of Ensemble-Based Models |
| title_short |
ICU Survival Prediction Incorporating Test-Time Augmentation to Improve the Accuracy of Ensemble-Based Models |
| title_full |
ICU Survival Prediction Incorporating Test-Time Augmentation to Improve the Accuracy of Ensemble-Based Models |
| title_fullStr |
ICU Survival Prediction Incorporating Test-Time Augmentation to Improve the Accuracy of Ensemble-Based Models |
| title_full_unstemmed |
ICU Survival Prediction Incorporating Test-Time Augmentation to Improve the Accuracy of Ensemble-Based Models |
| title_sort |
icu survival prediction incorporating test-time augmentation to improve the accuracy of ensemble-based models |
| publisher |
IEEE |
| series |
IEEE Access |
| issn |
2169-3536 |
| publishDate |
2021-01-01 |
| description |
This work presents a novel method for applying test-time augmentation (TTA) to tabular data. We used TTA along with an ensemble of 42 models to achieve higher performance on the MIT Global Open Source Severity of Illness Score dataset consisting of 131,051 ICU visits and outcomes. This method achieved an AUC of 0.915 on the private test set (19,669 admissions) and won first place at Stanford University’s WiDS Datathon 2020 challenge on Kaggle, while the Acute Physiology and Chronic Health Evaluation (APACHE) IV model (commonly used for ICU survival prediction in the literature) achieved an AUC of 0.868. In addition to increasing the AUC score, our method also reduces “unfair” bias. |
| topic |
Ensemble methods machine learning supervised classification healthcare |
| url |
https://ieeexplore.ieee.org/document/9462159/ |
| work_keys_str_mv |
AT sefficohen icusurvivalpredictionincorporatingtesttimeaugmentationtoimprovetheaccuracyofensemblebasedmodels AT noadagan icusurvivalpredictionincorporatingtesttimeaugmentationtoimprovetheaccuracyofensemblebasedmodels AT nuritcoheninger icusurvivalpredictionincorporatingtesttimeaugmentationtoimprovetheaccuracyofensemblebasedmodels AT danofer icusurvivalpredictionincorporatingtesttimeaugmentationtoimprovetheaccuracyofensemblebasedmodels AT liorrokach icusurvivalpredictionincorporatingtesttimeaugmentationtoimprovetheaccuracyofensemblebasedmodels |
| _version_ |
1721352463840182272 |