Predicting neurological outcome after out-of-hospital cardiac arrest with cumulative information; development and internal validation of an artificial neural network algorithm

Critical Care

Matt P. Wise¹  Jesper Kjaergaard²  Christian Hassager²  Attila Frigyesi³  Peder Andersson⁴  Hans Friberg⁵  Jesper Johnsson⁶  Niklas Nielsen⁶  Josef Dankiewicz⁷  Gisela Lilja⁸  Tobias Cronberg⁸  Johan Undén⁹  Ola Björnsson^1,10  Kaj Blennow^1,11  Henrik Zetterberg^1,12  Pascal Stammet^1,13 

[1] Adult Critical Care, University Hospital of Wales, Cardiff, UK;Department of Cardiology, Rigshospitalet and Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark;Department of Clinical Sciences Lund, Anaesthesia and Intensive Care, Lund University, Skåne University Hospital, Lund, Sweden;Centre for Mathematical Sciences, Mathematical Statistics, Lund University, Lund, Sweden;Department of Clinical Sciences Lund, Anaesthesia and Intensive Care, Lund University, Skåne University Hospital, Lund, Sweden;Department of Intensive and Perioperative Care, Skåne University Hospital, Getingevägen 4, 222 41, LundLund, Sweden;Department of Clinical Sciences Lund, Anaesthesia and Intensive Care, Lund University, Skåne University Hospital, Malmö, Sweden;Department of Clinical Sciences Lund, Anesthesia and Intensive Care, Lund University, Helsingborg Hospital, Lund, Sweden;Department of Clinical Sciences Lund, Cardiology, Lund University, Skåne University Hospital, Lund, Sweden;Department of Clinical Sciences Lund, Neurology, Lund University, Skåne University Hospital, Lund, Sweden;Department of Clinical Sciences Malmö, Anaesthesia and Intensive Care, Lund University, Hallands Hospital Halmstad, Halland, Sweden;Department of Energy Sciences, Faculty of Engineering, Lund University, Lund, Sweden;Centre for Mathematical Sciences, Mathematical Statistics, Lund University, Lund, Sweden;Department of Psychiatry and Neurochemistry, Institute of Neuroscience and Physiology, The Sahlgrenska Academy At the University of Gothenburg, Mölndal, Sweden;Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, Mölndal, Sweden;Department of Psychiatry and Neurochemistry, Institute of Neuroscience and Physiology, The Sahlgrenska Academy At the University of Gothenburg, Mölndal, Sweden;Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, Mölndal, Sweden;Department of Neurodegenerative Disease, UCL Institute of Neurology, Queen Square, London, UK;UK Dementia Research Institute At UCL, London, UK;Medical and Health Directorate, National Fire and Rescue Corps, 1, rue Robert Stumper, 2557, Luxembourg, Luxembourg;

关键词: Machine learning;    Artificial intelligence;    Artificial neural networks;    Neural networks;    Out-of-hospital cardiac arrest;    Cardiac arrest;    Cerebral performance category;    Critical care;    Intensive care;    Prediction;    Prognostication;   

DOI  :  10.1186/s13054-021-03505-9

来源: Springer

【 摘 要 】

BackgroundPrognostication of neurological outcome in patients who remain comatose after cardiac arrest resuscitation is complex. Clinical variables, as well as biomarkers of brain injury, cardiac injury, and systemic inflammation, all yield some prognostic value. We hypothesised that cumulative information obtained during the first three days of intensive care could produce a reliable model for predicting neurological outcome following out-of-hospital cardiac arrest (OHCA) using artificial neural network (ANN) with and without biomarkers.MethodsWe performed a post hoc analysis of 932 patients from the Target Temperature Management trial. We focused on comatose patients at 24, 48, and 72 h post-cardiac arrest and excluded patients who were awake or deceased at these time points. 80% of the patients were allocated for model development (training set) and 20% for internal validation (test set). To investigate the prognostic potential of different levels of biomarkers (clinically available and research-grade), patients’ background information, and intensive care observation and treatment, we created three models for each time point: (1) clinical variables, (2) adding clinically accessible biomarkers, e.g., neuron-specific enolase (NSE) and (3) adding research-grade biomarkers, e.g., neurofilament light (NFL). Patient outcome was the dichotomised Cerebral Performance Category (CPC) at six months; a good outcome was defined as CPC 1–2 whilst a poor outcome was defined as CPC 3–5. The area under the receiver operating characteristic curve (AUROC) was calculated for all test sets.ResultsAUROC remained below 90% when using only clinical variables throughout the first three days in the ICU. Adding clinically accessible biomarkers such as NSE, AUROC increased from 82 to 94% (p < 0.01). The prognostic accuracy remained excellent from day 1 to day 3 with an AUROC at approximately 95% when adding research-grade biomarkers. The models which included NSE after 72 h and NFL on any of the three days had a low risk of false-positive predictions while retaining a low number of false-negative predictions.ConclusionsIn this exploratory study, ANNs provided good to excellent prognostic accuracy in predicting neurological outcome in comatose patients post OHCA. The models which included NSE after 72 h and NFL on all days showed promising prognostic performance.

【 授权许可】

CC BY