Assessing the utility of a sliding-windows deep neural network approach for risk prediction of trauma patients

Research output: Contribution to journal › Journal article › Research › peer-review

Standard

Assessing the utility of a sliding-windows deep neural network approach for risk prediction of trauma patients. / Bonde, Alexander; Bonde, Mikkel; Troelsen, Anders; Sillesen, Martin.

In: Scientific Reports, Vol. 13, 5176, 2023.

Research output: Contribution to journal › Journal article › Research › peer-review

Harvard

Bonde, A, Bonde, M, Troelsen, A & Sillesen, M 2023, 'Assessing the utility of a sliding-windows deep neural network approach for risk prediction of trauma patients', Scientific Reports, vol. 13, 5176. https://doi.org/10.1038/s41598-023-32453-3

APA

Bonde, A., Bonde, M., Troelsen, A., & Sillesen, M. (2023). Assessing the utility of a sliding-windows deep neural network approach for risk prediction of trauma patients. Scientific Reports, 13, [5176]. https://doi.org/10.1038/s41598-023-32453-3

Vancouver

Bonde A, Bonde M, Troelsen A, Sillesen M. Assessing the utility of a sliding-windows deep neural network approach for risk prediction of trauma patients. Scientific Reports. 2023;13. 5176. https://doi.org/10.1038/s41598-023-32453-3

Author

Bonde, Alexander ; Bonde, Mikkel ; Troelsen, Anders ; Sillesen, Martin. / Assessing the utility of a sliding-windows deep neural network approach for risk prediction of trauma patients. In: Scientific Reports. 2023 ; Vol. 13.

Bibtex

@article{dbb22044356843e78e9270a349d47c31,

title = "Assessing the utility of a sliding-windows deep neural network approach for risk prediction of trauma patients",

abstract = "The risks of post trauma complications are regulated by the injury, comorbidities, and the clinical trajectories, yet prediction models are often limited to single time-point data. We hypothesize that deep learning prediction models can be used for risk prediction using additive data after trauma using a sliding windows approach. Using the American College of Surgeons Trauma Quality Improvement Program (ACS TQIP) database, we developed three deep neural network models, for sliding-windows risk prediction. Output variables included early- and late mortality and any of 17 complications. As patients moved through the treatment trajectories, performance metrics increased. Models predicted early- and late mortality with ROC AUCs ranging from 0.980 to 0.994 and 0.910 to 0.972, respectively. For the remaining 17 complications, the mean performance ranged from 0.829 to 0.912. In summary, the deep neural networks achieved excellent performance in the sliding windows risk stratification of trauma patients.",

author = "Alexander Bonde and Mikkel Bonde and Anders Troelsen and Martin Sillesen",

note = "Publisher Copyright: {\textcopyright} 2023, The Author(s).",

year = "2023",

doi = "10.1038/s41598-023-32453-3",

language = "English",

volume = "13",

journal = "Scientific Reports",

issn = "2045-2322",

publisher = "nature publishing group",

}

RIS

TY - JOUR

T1 - Assessing the utility of a sliding-windows deep neural network approach for risk prediction of trauma patients

AU - Bonde, Alexander

AU - Bonde, Mikkel

AU - Troelsen, Anders

AU - Sillesen, Martin

PY - 2023

Y1 - 2023

N2 - The risks of post trauma complications are regulated by the injury, comorbidities, and the clinical trajectories, yet prediction models are often limited to single time-point data. We hypothesize that deep learning prediction models can be used for risk prediction using additive data after trauma using a sliding windows approach. Using the American College of Surgeons Trauma Quality Improvement Program (ACS TQIP) database, we developed three deep neural network models, for sliding-windows risk prediction. Output variables included early- and late mortality and any of 17 complications. As patients moved through the treatment trajectories, performance metrics increased. Models predicted early- and late mortality with ROC AUCs ranging from 0.980 to 0.994 and 0.910 to 0.972, respectively. For the remaining 17 complications, the mean performance ranged from 0.829 to 0.912. In summary, the deep neural networks achieved excellent performance in the sliding windows risk stratification of trauma patients.

AB - The risks of post trauma complications are regulated by the injury, comorbidities, and the clinical trajectories, yet prediction models are often limited to single time-point data. We hypothesize that deep learning prediction models can be used for risk prediction using additive data after trauma using a sliding windows approach. Using the American College of Surgeons Trauma Quality Improvement Program (ACS TQIP) database, we developed three deep neural network models, for sliding-windows risk prediction. Output variables included early- and late mortality and any of 17 complications. As patients moved through the treatment trajectories, performance metrics increased. Models predicted early- and late mortality with ROC AUCs ranging from 0.980 to 0.994 and 0.910 to 0.972, respectively. For the remaining 17 complications, the mean performance ranged from 0.829 to 0.912. In summary, the deep neural networks achieved excellent performance in the sliding windows risk stratification of trauma patients.

U2 - 10.1038/s41598-023-32453-3

DO - 10.1038/s41598-023-32453-3

M3 - Journal article

C2 - 36997598

AN - SCOPUS:85151316197

VL - 13

JO - Scientific Reports

JF - Scientific Reports

SN - 2045-2322

M1 - 5176

ER -

ID: 344447655