AUTHOR=Baur David , Gehlen Tobias , Scherer Julian , Back David Alexander , Tsitsilonis Serafeim , Kabir Koroush , Osterhoff Georg 

TITLE=Decision support by machine learning systems for acute management of severely injured patients: A systematic review

JOURNAL=Frontiers in Surgery

VOLUME=Volume 9 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/surgery/articles/10.3389/fsurg.2022.924810

DOI=10.3389/fsurg.2022.924810

ISSN=2296-875X

ABSTRACT=Introduction 
The treatment of severely injured patients requires numerous critical decisions within short time intervals in a highly complex situation. The coordination of a trauma team in this setting has been shown to be associated with multiple procedural errors even in experienced care teams.  Machine learning (ML) is an approach that estimates outcomes based on past experiences and data patterns using a computer-generated algorithm. This systematic review aimed to summarize the existing literature on the value of ML for the initial management of severely injured patients. 
Methods 
We conducted a systematic review of the literature with the goal of finding all articles describing the use of ML systems in the context of acute management of severely injured patients. MESH searches of Pubmed/Medline and Web of science were conducted. Studies with number of inclusion of less than 10 patients were excluded. Studies were divided into following main prediction groups: 1) injury pattern, 2) haemorrhage / need for transfusion, 3) emergency intervention, 4) ICU / length of hospital stay, 5) mortality.
Results 
Thirty-six articles met the inclusion criteria, among these were two prospective and 34 retrospective case series. Publication dates ranged from 2000 to 2020 and included 32 different first-authors. A total of 18,586,929 patients were included into the prediction models. Mortality was the most represented main prediction group (n=19). ML models used were artificial neural networks (ANN, n = 15), singular vector machines (n = 3), Bayesian networks (n = 7), random forest (n = 6), natural language processing (n = 2), stacked ensemble classifiers (SuperLearner (SL), n = 3), k-nearest neighbor (KNN, n = 1), belief systems (n = 1) and sequential minimal optimization (n = 2). Thirty articles assessed results as positive, whilst five showed moderate results, and one article described negative results to their implementation of the respective prediction model. 
Conclusions 
Whilst the majority of articles show a generally positive result with high accuracy and precision, there are several requirements that need to be met, to make implementation of such models in the daily clinical work possible.