Refining outcome prediction after traumatic brain injury with machine learning algorithms
Abstract Outcome after traumatic brain injury (TBI) is typically assessed using the Glasgow outcome scale extended (GOSE) with levels from 1 (death) to 8 (upper good recovery). Outcome prediction has classically been dichotomized into either dead/alive or favorable/unfavorable outcome. Binary outcom...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2024-04-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-024-58527-4 |
| _version_ | 1797219808778911744 |
|---|---|
| author | D. Bark M. Boman B. Depreitere D. W. Wright A. Lewén P. Enblad A. Hånell E. Rostami |
| author_facet | D. Bark M. Boman B. Depreitere D. W. Wright A. Lewén P. Enblad A. Hånell E. Rostami |
| author_sort | D. Bark |
| collection | DOAJ |
| description | Abstract Outcome after traumatic brain injury (TBI) is typically assessed using the Glasgow outcome scale extended (GOSE) with levels from 1 (death) to 8 (upper good recovery). Outcome prediction has classically been dichotomized into either dead/alive or favorable/unfavorable outcome. Binary outcome prediction models limit the possibility of detecting subtle yet significant improvements. We set out to explore different machine learning methods with the purpose of mapping their predictions to the full 8 grade scale GOSE following TBI. The models were set up using the variables: age, GCS-motor score, pupillary reaction, and Marshall CT score. For model setup and internal validation, a total of 866 patients could be included. For external validation, a cohort of 369 patients were included from Leuven, Belgium, and a cohort of 573 patients from the US multi-center ProTECT III study. Our findings indicate that proportional odds logistic regression (POLR), random forest regression, and a neural network model achieved accuracy values of 0.3–0.35 when applied to internal data, compared to the random baseline which is 0.125 for eight categories. The models demonstrated satisfactory performance during external validation in the data from Leuven, however, their performance were not satisfactory when applied to the ProTECT III dataset. |
| first_indexed | 2024-04-24T12:39:32Z |
| format | Article |
| id | doaj.art-8c327a627efa4cb4a5a2c3f3dcc789f4 |
| institution | Directory Open Access Journal |
| issn | 2045-2322 |
| language | English |
| last_indexed | 2024-04-24T12:39:32Z |
| publishDate | 2024-04-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj.art-8c327a627efa4cb4a5a2c3f3dcc789f42024-04-07T11:19:35ZengNature PortfolioScientific Reports2045-23222024-04-0114111510.1038/s41598-024-58527-4Refining outcome prediction after traumatic brain injury with machine learning algorithmsD. Bark0M. Boman1B. Depreitere2D. W. Wright3A. Lewén4P. Enblad5A. Hånell6E. Rostami7Department of Medical Sciences Neurosurgery, Uppsala UniversityDivision of Clinical Epidemiology, Department of Medicine SolnaDepartment of Neurosurgery, University Hospitals LeuvenDepartment of Emergency Medicine, Emory UniversityDepartment of Medical Sciences Neurosurgery, Uppsala UniversityDepartment of Medical Sciences Neurosurgery, Uppsala UniversityDepartment of Medical Sciences Neurosurgery, Uppsala UniversityDepartment of Medical Sciences Neurosurgery, Uppsala UniversityAbstract Outcome after traumatic brain injury (TBI) is typically assessed using the Glasgow outcome scale extended (GOSE) with levels from 1 (death) to 8 (upper good recovery). Outcome prediction has classically been dichotomized into either dead/alive or favorable/unfavorable outcome. Binary outcome prediction models limit the possibility of detecting subtle yet significant improvements. We set out to explore different machine learning methods with the purpose of mapping their predictions to the full 8 grade scale GOSE following TBI. The models were set up using the variables: age, GCS-motor score, pupillary reaction, and Marshall CT score. For model setup and internal validation, a total of 866 patients could be included. For external validation, a cohort of 369 patients were included from Leuven, Belgium, and a cohort of 573 patients from the US multi-center ProTECT III study. Our findings indicate that proportional odds logistic regression (POLR), random forest regression, and a neural network model achieved accuracy values of 0.3–0.35 when applied to internal data, compared to the random baseline which is 0.125 for eight categories. The models demonstrated satisfactory performance during external validation in the data from Leuven, however, their performance were not satisfactory when applied to the ProTECT III dataset.https://doi.org/10.1038/s41598-024-58527-4 |
| spellingShingle | D. Bark M. Boman B. Depreitere D. W. Wright A. Lewén P. Enblad A. Hånell E. Rostami Refining outcome prediction after traumatic brain injury with machine learning algorithms Scientific Reports |
| title | Refining outcome prediction after traumatic brain injury with machine learning algorithms |
| title_full | Refining outcome prediction after traumatic brain injury with machine learning algorithms |
| title_fullStr | Refining outcome prediction after traumatic brain injury with machine learning algorithms |
| title_full_unstemmed | Refining outcome prediction after traumatic brain injury with machine learning algorithms |
| title_short | Refining outcome prediction after traumatic brain injury with machine learning algorithms |
| title_sort | refining outcome prediction after traumatic brain injury with machine learning algorithms |
| url | https://doi.org/10.1038/s41598-024-58527-4 |
| work_keys_str_mv | AT dbark refiningoutcomepredictionaftertraumaticbraininjurywithmachinelearningalgorithms AT mboman refiningoutcomepredictionaftertraumaticbraininjurywithmachinelearningalgorithms AT bdepreitere refiningoutcomepredictionaftertraumaticbraininjurywithmachinelearningalgorithms AT dwwright refiningoutcomepredictionaftertraumaticbraininjurywithmachinelearningalgorithms AT alewen refiningoutcomepredictionaftertraumaticbraininjurywithmachinelearningalgorithms AT penblad refiningoutcomepredictionaftertraumaticbraininjurywithmachinelearningalgorithms AT ahanell refiningoutcomepredictionaftertraumaticbraininjurywithmachinelearningalgorithms AT erostami refiningoutcomepredictionaftertraumaticbraininjurywithmachinelearningalgorithms |