TY - JOUR
T1 - Decision support system and outcome prediction in a cohort of patients with necrotizing soft-tissue infections
AU - Katz, Sonja
AU - Suijker, Jaco
AU - Hardt, Christopher
AU - Madsen, Martin Bruun
AU - Vries, Annebeth Meij-de
AU - Pijpe, Anouk
AU - Skrede, Steinar
AU - Hyldegaard, Ole
AU - Solligård, Erik
AU - Norrby-Teglund, Anna
AU - Saccenti, Edoardo
AU - Martins dos Santos, Vitor A. P.
N1 - Funding Information: The authors are grateful to all members of the INFECT, PerAID and PerMIT collaborations, as extensive discussion within the consortium have greatly contributed to the finalisation of this manuscript. Beside named authors in this article, the INFECT Study Group includes: Michael Nekludov, Ylva Karlsson, Per Arnell, Morten Hedetoft, Marco B. Hansen, Peter Polzik, Daniel Bidstrup, Nina F. Bærnthsen, Gladis H. Frendø, Erik C. Jansen, Lærke B. Madsen, Rasmus B. Müller, Emilie M. J. Pedersen, Marie W. Petersen, Frederikke Ravn, Isabel F. G. Smidt-Nielsen, Anna M. Wahl, Sandra Wulffeld, Sara Aronsson, Anders Rosemar, Joakim Trogen, Trond Bruun, Torbjørn Nedrebø, Oddvar Oppegaard, Eivind Rath and Marianne Sævik. The PerAID/PerMIT Study groups, besides named authors, include: Mattias Svensson, Kristoffer Strålin, Trond Bruun, Oddvar Oppegaard, Knut Anders Mosevoll, Jan Kristian Damås, Paul van Zuijlen, Laura M. Palma Medina, Lorna Morris, and Marco Anteghini. The full list of all Study Group members, national site investigators, and their affiliation can be found in Supplementary Note 1. The authors are indebted to Stephan G.F. Papendorp, Marieke Verhaar, Fabienne A.M. Roossien, Evelien de Jong and Vincent M. de Jong for their participation in the interviews. The authors thank the supporters of this study: the European Union Seventh Framework Programme (FP7/2007–2013) under the grant agreement 305340 (INFECT project); the Swedish Governmental Agency for Innovation Systems (VINNOVA), Innovation Fund Denmark, and the Research Council of Norway under the frame of NordForsk (project No. 90456, PerAID); the Swedish Research Council, Innovation Fund Denmark, the Research Council of Norway, the Netherlands Organisation for Health Research and Development (ZonMW), and DLR Federal Ministry of Education and Research, under the frame of ERA PerMed (project 2018–151, PerMIT); the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 860895 TranSYS. Funding Information: The authors are grateful to all members of the INFECT, PerAID and PerMIT collaborations, as extensive discussion within the consortium have greatly contributed to the finalisation of this manuscript. Beside named authors in this article, the INFECT Study Group includes: Michael Nekludov, Ylva Karlsson, Per Arnell, Morten Hedetoft, Marco B. Hansen, Peter Polzik, Daniel Bidstrup, Nina F. Bærnthsen, Gladis H. Frendø, Erik C. Jansen, Lærke B. Madsen, Rasmus B. Müller, Emilie M. J. Pedersen, Marie W. Petersen, Frederikke Ravn, Isabel F. G. Smidt-Nielsen, Anna M. Wahl, Sandra Wulffeld, Sara Aronsson, Anders Rosemar, Joakim Trogen, Trond Bruun, Torbjørn Nedrebø, Oddvar Oppegaard, Eivind Rath and Marianne Sævik. The PerAID/PerMIT Study groups, besides named authors, include: Mattias Svensson, Kristoffer Strålin, Trond Bruun, Oddvar Oppegaard, Knut Anders Mosevoll, Jan Kristian Damås, Paul van Zuijlen, Laura M. Palma Medina, Lorna Morris, and Marco Anteghini. The full list of all Study Group members, national site investigators, and their affiliation can be found in Supplementary Note 1. The authors are indebted to Stephan G.F. Papendorp, Marieke Verhaar, Fabienne A.M. Roossien, Evelien de Jong and Vincent M. de Jong for their participation in the interviews. The authors thank the supporters of this study: the European Union Seventh Framework Programme (FP7/2007–2013) under the grant agreement 305340 (INFECT project); the Swedish Governmental Agency for Innovation Systems (VINNOVA), Innovation Fund Denmark, and the Research Council of Norway under the frame of NordForsk (project No. 90456, PerAID); the Swedish Research Council, Innovation Fund Denmark, the Research Council of Norway, the Netherlands Organisation for Health Research and Development (ZonMW), and DLR Federal Ministry of Education and Research, under the frame of ERA PerMed (project 2018–151, PerMIT); the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 860895 TranSYS. Publisher Copyright: © 2022 The Authors
PY - 2022/11/1
Y1 - 2022/11/1
N2 - Introduction: Necrotizing Soft Tissue Infections (NSTI) are severe infections with high mortality affecting a heterogeneous patient population. There is a need for a clinical decision support system which predicts outcomes and provides treatment recommendations early in the disease course. Methods: To identify relevant clinical needs, interviews with eight medical professionals (surgeons, intensivists, general practitioner, emergency department physician) were conducted. This resulted in 24 unique questions. Mortality was selected as first endpoint to develop a machine learning (Random Forest) based prediction model. For this purpose, data from the prospective, international INFECT cohort (N = 409) was used. Results: Applying a feature selection procedure based on an unsupervised algorithm (Boruta) to the > 1000 variables available in INFECT, including baseline, and both NSTI specific and NSTI non-specific clinical data yielded sixteen predictive parameters available on or prior to the first day on the intensive care unit (ICU). Using these sixteen variables 30-day mortality could be accurately predicted (AUC = 0.91, 95% CI 0.88–0.96). Except for age, all variables were related to sepsis (e.g. lactate, urine production, systole). No NSTI-specific variables were identified. Predictions significantly outperformed the SOFA score(p < 0.001, AUC = 0.77, 95% CI 0.69–0.84) and exceeded but did not significantly differ from the SAPS II score (p = 0.07, AUC = 0.88, 95% CI 0.83–0.92). The developed model proved to be stable with AUC > 0.8 in case of high rates of missing data (50% missing) or when only using very early (<1 h) available variables. Conclusions: This study shows that mortality can be accurately predicted using a machine learning model. It lays the foundation for a more extensive, multi-endpoint clinical decision support system in which ultimately other outcomes and clinical questions (risk for septic shock, AKI, causative microbe) will be included.
AB - Introduction: Necrotizing Soft Tissue Infections (NSTI) are severe infections with high mortality affecting a heterogeneous patient population. There is a need for a clinical decision support system which predicts outcomes and provides treatment recommendations early in the disease course. Methods: To identify relevant clinical needs, interviews with eight medical professionals (surgeons, intensivists, general practitioner, emergency department physician) were conducted. This resulted in 24 unique questions. Mortality was selected as first endpoint to develop a machine learning (Random Forest) based prediction model. For this purpose, data from the prospective, international INFECT cohort (N = 409) was used. Results: Applying a feature selection procedure based on an unsupervised algorithm (Boruta) to the > 1000 variables available in INFECT, including baseline, and both NSTI specific and NSTI non-specific clinical data yielded sixteen predictive parameters available on or prior to the first day on the intensive care unit (ICU). Using these sixteen variables 30-day mortality could be accurately predicted (AUC = 0.91, 95% CI 0.88–0.96). Except for age, all variables were related to sepsis (e.g. lactate, urine production, systole). No NSTI-specific variables were identified. Predictions significantly outperformed the SOFA score(p < 0.001, AUC = 0.77, 95% CI 0.69–0.84) and exceeded but did not significantly differ from the SAPS II score (p = 0.07, AUC = 0.88, 95% CI 0.83–0.92). The developed model proved to be stable with AUC > 0.8 in case of high rates of missing data (50% missing) or when only using very early (<1 h) available variables. Conclusions: This study shows that mortality can be accurately predicted using a machine learning model. It lays the foundation for a more extensive, multi-endpoint clinical decision support system in which ultimately other outcomes and clinical questions (risk for septic shock, AKI, causative microbe) will be included.
KW - Clinical decision support system
KW - Intensive care unit
KW - Machine learning
KW - Mortality
KW - Necrotizing soft-tissue infections
KW - Random forest
UR - http://www.scopus.com/inward/record.url?scp=85138993689&partnerID=8YFLogxK
U2 - https://doi.org/10.1016/j.ijmedinf.2022.104878
DO - https://doi.org/10.1016/j.ijmedinf.2022.104878
M3 - Article
C2 - 36194993
SN - 1386-5056
VL - 167
JO - International Journal of Medical Informatics
JF - International Journal of Medical Informatics
M1 - 104878
ER -