Pedro Kurtz1,2,3, Igor Tona Peres4, Marcio Soares1, Jorge I F Salluh1,5, Fernando A Bozza6,7. 1. D'Or Institute for Research and Education, Rio de Janeiro, Brazil. 2. Hospital Copa Star, Rio de Janeiro, Brazil. 3. Paulo Niemeyer State Brain Institute, Rio de Janeiro, Brazil. 4. Department of Industrial Engineering, Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil. 5. Postgraduate Program of Internal Medicine, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil. 6. D'Or Institute for Research and Education, Rio de Janeiro, Brazil. bozza.fernando@gmail.com. 7. National Institute of Infectious Disease Evandro Chagas, Oswaldo Cruz Foundation, Rio de Janeiro, Brazil. bozza.fernando@gmail.com.
Abstract
BACKGROUND: Hospital length of stay and mortality are associated with resource use and clinical severity, respectively, in patients admitted to the intensive care unit (ICU) with acute stroke. We proposed a structured data-driven methodology to develop length of stay and 30-day mortality prediction models in a large multicenter Brazilian ICU cohort. METHODS: We analyzed data from 130 ICUs from 43 Brazilian hospitals. All consecutive adult patients admitted with stroke (ischemic or nontraumatic hemorrhagic) to the ICU from January 2011 to December 2020 were included. Demographic data, comorbidities, acute disease characteristics, organ support, and laboratory data were retrospectively analyzed by a data-driven methodology, which included seven different types of machine learning models applied to training and test sets of data. The best performing models, based on discrimination and calibration measures, are reported as the main results. Outcomes were hospital length of stay and 30-day in-hospital mortality. RESULTS: Of 17,115 ICU admissions for stroke, 16,592 adult patients (13,258 ischemic and 3334 hemorrhagic) were analyzed; 4298 (26%) patients had a prolonged hospital length of stay (> 14 days), and 30-day mortality was 8% (n = 1392). Prolonged hospital length of stay was best predicted by the random forests model (Brier score = 0.17, area under the curve = 0.73, positive predictive value = 0.61, negative predictive value = 0.78). Mortality prediction also yielded the best discrimination and calibration through random forests (Brier score = 0.05, area under the curve = 0.90, positive predictive value = 0.66, negative predictive value = 0.94). Among the 20 strongest contributor variables in both models were (1) premorbid conditions (e.g., functional impairment), (2) multiple organ dysfunction parameters (e.g., hypotension, mechanical ventilation), and (3) acute neurological aspects of stroke (e.g., Glasgow coma scale score on admission, stroke type). CONCLUSIONS: Hospital length of stay and 30-day mortality of patients admitted to the ICU with stroke were accurately predicted through machine learning methods, even in the absence of stroke-specific data, such as the National Institutes of Health Stroke Scale score or neuroimaging findings. The proposed methods using general intensive care databases may be used for resource use allocation planning and performance assessment of ICUs treating stroke. More detailed acute neurological and management data, as well as long-term functional outcomes, may improve the accuracy and applicability of future machine-learning-based prediction algorithms.
BACKGROUND: Hospital length of stay and mortality are associated with resource use and clinical severity, respectively, in patients admitted to the intensive care unit (ICU) with acute stroke. We proposed a structured data-driven methodology to develop length of stay and 30-day mortality prediction models in a large multicenter Brazilian ICU cohort. METHODS: We analyzed data from 130 ICUs from 43 Brazilian hospitals. All consecutive adult patients admitted with stroke (ischemic or nontraumatic hemorrhagic) to the ICU from January 2011 to December 2020 were included. Demographic data, comorbidities, acute disease characteristics, organ support, and laboratory data were retrospectively analyzed by a data-driven methodology, which included seven different types of machine learning models applied to training and test sets of data. The best performing models, based on discrimination and calibration measures, are reported as the main results. Outcomes were hospital length of stay and 30-day in-hospital mortality. RESULTS: Of 17,115 ICU admissions for stroke, 16,592 adult patients (13,258 ischemic and 3334 hemorrhagic) were analyzed; 4298 (26%) patients had a prolonged hospital length of stay (> 14 days), and 30-day mortality was 8% (n = 1392). Prolonged hospital length of stay was best predicted by the random forests model (Brier score = 0.17, area under the curve = 0.73, positive predictive value = 0.61, negative predictive value = 0.78). Mortality prediction also yielded the best discrimination and calibration through random forests (Brier score = 0.05, area under the curve = 0.90, positive predictive value = 0.66, negative predictive value = 0.94). Among the 20 strongest contributor variables in both models were (1) premorbid conditions (e.g., functional impairment), (2) multiple organ dysfunction parameters (e.g., hypotension, mechanical ventilation), and (3) acute neurological aspects of stroke (e.g., Glasgow coma scale score on admission, stroke type). CONCLUSIONS: Hospital length of stay and 30-day mortality of patients admitted to the ICU with stroke were accurately predicted through machine learning methods, even in the absence of stroke-specific data, such as the National Institutes of Health Stroke Scale score or neuroimaging findings. The proposed methods using general intensive care databases may be used for resource use allocation planning and performance assessment of ICUs treating stroke. More detailed acute neurological and management data, as well as long-term functional outcomes, may improve the accuracy and applicability of future machine-learning-based prediction algorithms.
Authors: Stephen Bacchi; Luke Oakden-Rayner; David K Menon; Jim Jannes; Timothy Kleinig; Simon Koblar Journal: J Clin Neurosci Date: 2020-08-05 Impact factor: 1.961
Authors: Sheila C Ouriques Martins; Claudio Sacks; Werner Hacke; Michael Brainin; Francisco de Assis Figueiredo; Octávio Marques Pontes-Neto; Pablo M Lavados Germain; Maria F Marinho; Arnold Hoppe Wiegering; Diana Vaca McGhie; Salvador Cruz-Flores; Sebastian F Ameriso; Walter M Camargo Villareal; Juan Carlos Durán; José E Fogolin Passos; Raul Gomes Nogueira; João J Freitas de Carvalho; Gisele Sampaio Silva; Carla H Cabral Moro; Jamary Oliveira-Filho; Rubens Gagliardi; Eduardo D Gomes de Sousa; Felipe Fagundes Soares; Katia de Pinho Campos; Paulo F Piza Teixeira; Ivete Pillo Gonçalves; Irving R Santos Carquin; Mário Muñoz Collazos; Germán E Pérez Romero; Javier I Maldonado Figueredo; Miguel A Barboza; Miguel Á Celis López; Fernando Góngora-Rivera; Carlos Cantú-Brito; Nelson Novarro-Escudero; Miguel Á Velázquez Blanco; Carlos A Arbo Oze de Morvil; Aurora B Olmedo Bareiro; Gloria Meza Rojas; Alan Flores; Jorge Arturo Hancco-Saavedra; Vivian Pérez Jimenez; Carlos Abanto Argomedo; Liliana Rodriguez Kadota; Roberto Crosa; Daissy L Mora Cuervo; Ana C de Souza; Leonardo A Carbonera; Tony F Álvarez Guzmán; Nelson Maldonado; Norberto L Cabral; Craig Anderson; Patrice Lindsay; Anselm Hennis; Valery L Feigin Journal: Lancet Neurol Date: 2019-04-24 Impact factor: 44.182
Authors: Carlos Fernandez-Lozano; Pablo Hervella; Virginia Mato-Abad; Manuel Rodríguez-Yáñez; Sonia Suárez-Garaboa; Iria López-Dequidt; Ana Estany-Gestal; Tomás Sobrino; Francisco Campos; José Castillo; Santiago Rodríguez-Yáñez; Ramón Iglesias-Rey Journal: Sci Rep Date: 2021-05-12 Impact factor: 4.379
Authors: Leila F Dantas; Janaina F Marchesi; Igor T Peres; Silvio Hamacher; Fernando A Bozza; Ricardo A Quintano Neira Journal: PLoS One Date: 2019-03-19 Impact factor: 3.240