Zina M Ibrahim1,2,3, Honghan Wu4, Ahmed Hamoud5, Lukas Stappen6, Richard J B Dobson1,2,3, Andrea Agarossi7. 1. Department of Biostatistics & Health Informatics, King's College London, London, UK. 2. Institute of Health Informatics, University College London, London, UK. 3. Health Data Research UK, University College London, London, UK. 4. Usher Institute of Population Health Sciences and Informatics, University of Edinburgh, Edinburgh, UK. 5. Department of Renal Medicine, East and North Hertfordshire NHS Trust, Stevenage, UK. 6. Embedded Intelligence for Health Care and Wellbeing, University of Augsburg, Augsburg, Germany. 7. Department of Anaesthesia and Intensive Care, Luigi Sacco Hospital, Milan, Italy.
Abstract
OBJECTIVES: Current machine learning models aiming to predict sepsis from electronic health records (EHR) do not account 20 for the heterogeneity of the condition despite its emerging importance in prognosis and treatment. This work demonstrates the added value of stratifying the types of organ dysfunction observed in patients who develop sepsis in the intensive care unit (ICU) in improving the ability to recognize patients at risk of sepsis from their EHR data. MATERIALS AND METHODS: Using an ICU dataset of 13 728 records, we identify clinically significant sepsis subpopulations with distinct organ dysfunction patterns. We perform classification experiments with random forest, gradient boost trees, and support vector machines, using the identified subpopulations to distinguish patients who develop sepsis in the ICU from those who do not. RESULTS: The classification results show that features selected using sepsis subpopulations as background knowledge yield a superior performance in distinguishing septic from non-septic patients regardless of the classification model used. The improved performance is especially pronounced in specificity, which is a current bottleneck in sepsis prediction machine learning models. CONCLUSION: Our findings can steer machine learning efforts toward more personalized models for complex conditions including sepsis.
OBJECTIVES: Current machine learning models aiming to predict sepsis from electronic health records (EHR) do not account 20 for the heterogeneity of the condition despite its emerging importance in prognosis and treatment. This work demonstrates the added value of stratifying the types of organ dysfunction observed in patients who develop sepsis in the intensive care unit (ICU) in improving the ability to recognize patients at risk of sepsis from their EHR data. MATERIALS AND METHODS: Using an ICU dataset of 13 728 records, we identify clinically significant sepsis subpopulations with distinct organ dysfunction patterns. We perform classification experiments with random forest, gradient boost trees, and support vector machines, using the identified subpopulations to distinguish patients who develop sepsis in the ICU from those who do not. RESULTS: The classification results show that features selected using sepsis subpopulations as background knowledge yield a superior performance in distinguishing septic from non-septic patients regardless of the classification model used. The improved performance is especially pronounced in specificity, which is a current bottleneck in sepsis prediction machine learning models. CONCLUSION: Our findings can steer machine learning efforts toward more personalized models for complex conditions including sepsis.
Authors: Evelyn M Olenick; Kathie S Zimbro; Gabrielle M DʼLima; Patricia Ver Schneider; Danielle Jones Journal: J Nurs Care Qual Date: 2017 Jan/Mar Impact factor: 1.597
Authors: Daniel B Knox; Michael J Lanspa; Kathryn G Kuttler; Simon C Brewer; Samuel M Brown Journal: Intensive Care Med Date: 2015-04-08 Impact factor: 17.440
Authors: Michael Rothman; Mitchell Levy; R Philip Dellinger; Stephen L Jones; Robert L Fogerty; Kirk G Voelker; Barry Gross; Albert Marchetti; Joseph Beals Journal: J Crit Care Date: 2016-12-03 Impact factor: 3.425
Authors: Åsa Askim; Florentin Moser; Lise T Gustad; Helga Stene; Maren Gundersen; Bjørn Olav Åsvold; Jostein Dale; Lars Petter Bjørnsen; Jan Kristian Damås; Erik Solligård Journal: Scand J Trauma Resusc Emerg Med Date: 2017-06-09 Impact factor: 2.953
Authors: Alistair E W Johnson; Tom J Pollard; Lu Shen; Li-Wei H Lehman; Mengling Feng; Mohammad Ghassemi; Benjamin Moody; Peter Szolovits; Leo Anthony Celi; Roger G Mark Journal: Sci Data Date: 2016-05-24 Impact factor: 6.444
Authors: Muge Capan; Stephen Hoover; Kristen E Miller; Carmen Pal; Justin M Glasgow; Eric V Jackson; Ryan C Arnold Journal: BMJ Open Qual Date: 2018-08-10
Authors: Edward J Schenck; Katherine L Hoffman; Marika Cusick; Joseph Kabariti; Evan T Sholle; Thomas R Campion Journal: J Biomed Inform Date: 2021-04-14 Impact factor: 8.000
Authors: Daniele Roberto Giacobbe; Alessio Signori; Filippo Del Puente; Sara Mora; Luca Carmisciano; Federica Briano; Antonio Vena; Lorenzo Ball; Chiara Robba; Paolo Pelosi; Mauro Giacomini; Matteo Bassetti Journal: Front Med (Lausanne) Date: 2021-02-12
Authors: Marc P Maurits; Ilya Korsunsky; Soumya Raychaudhuri; Shawn N Murphy; Jordan W Smoller; Scott T Weiss; Thomas W J Huizinga; Marcel J T Reinders; Elizabeth W Karlson; Erik B van den Akker; Rachel Knevel Journal: J Am Med Inform Assoc Date: 2022-04-13 Impact factor: 7.942