Journal of Clinical Monitoring and Computing

, Volume 33, Issue 1, pp 39–51 | Cite as

Forewarning of hypotensive events using a Bayesian artificial neural network in neurocritical care

  • Rob Donald
  • Tim Howells
  • Ian PiperEmail author
  • P. Enblad
  • P. Nilsson
  • I. Chambers
  • B. Gregson
  • G. Citerio
  • K. Kiening
  • J. Neumann
  • A. Ragauskas
  • J. Sahuquillo
  • R. Sinnott
  • A. Stell
  • the BrainIT Group
Original Research


Traumatically brain injured (TBI) patients are at risk from secondary insults. Arterial hypotension, critically low blood pressure, is one of the most dangerous secondary insults and is related to poor outcome in patients. The overall aim of this study was to get proof of the concept that advanced statistical techniques (machine learning) are methods that are able to provide early warning of impending hypotensive events before they occur during neuro-critical care. A Bayesian artificial neural network (BANN) model predicting episodes of hypotension was developed using data from 104 patients selected from the BrainIT multi-center database. Arterial hypotension events were recorded and defined using the Edinburgh University Secondary Insult Grades (EUSIG) physiological adverse event scoring system. The BANN was trained on a random selection of 50% of the available patients (n = 52) and validated on the remaining cohort. A multi-center prospective pilot study (Phase 1, n = 30) was then conducted with the system running live in the clinical environment, followed by a second validation pilot study (Phase 2, n = 49). From these prospectively collected data, a final evaluation study was done on 69 of these patients with 10 patients excluded from the Phase 2 study because of insufficient or invalid data. Each data collection phase was a prospective non-interventional observational study conducted in a live clinical setting to test the data collection systems and the model performance. No prediction information was available to the clinical teams during a patient’s stay in the ICU. The final cohort (n = 69), using a decision threshold of 0.4, and including false positive checks, gave a sensitivity of 39.3% (95% CI 32.9–46.1) and a specificity of 91.5% (95% CI 89.0–93.7). Using a decision threshold of 0.3, and false positive correction, gave a sensitivity of 46.6% (95% CI 40.1–53.2) and specificity of 85.6% (95% CI 82.3–88.8). With a decision threshold of 0.3, > 15 min warning of patient instability can be achieved. We have shown, using advanced machine learning techniques running in a live neuro-critical care environment, that it would be possible to give neurointensive teams early warning of potential hypotensive events before they emerge, allowing closer monitoring and earlier clinical assessment in an attempt to prevent the onset of hypotension. The multi-centre clinical infrastructure developed to support the clinical studies provides a solid base for further collaborative research on data quality, false positive correction and the display of early warning data in a clinical setting.


Traumatic brain injury Neuro-intensive care Bayesian prediction Clinical study results 



Funding: European 2020 IST Programme - IST-2007-217049; Investigators and participating centers: Barcelona, Spain: Prof. Sahuquillo; Cambridge, UK: Prof. Pickard; Edinburgh, UK: Prof. Whittle; Glasgow, UK: Mr. Dunn; Gothenburg, Sweden: Dr. Rydenhag; Heidelberg, Germany: Dr. Kiening; Iasi, Romania: Dr. Iencean; Kaunas, Lithuania: Prof. Pavalkis; Leipzig, Germany: Prof. Meixensberger; Leuven, Belgium: Prof. Goffin; Mannheim, Germany: Prof. Vajkoczy; Milano, Italy: Prof. Stocchetti; Monza, Italy: Dr. Citerio; Newcastle upon Tyne, UK: Dr. Chambers; Novara, Italy: Prof. Della Corte; Southampton, UK: Dr. Hell; Uppsala, Sweden: Prof. Enblad; Torino, Italy: Dr. Mascia; Vilnius, Lithuania: Prof. Jarzemaskas; Zu¨rich, Switzerland: Prof. Stocker.

BrainIT Steering Group members: IR Chambers, Newcastle upon Tyne; G Citerio, Monza; P Enblad, Uppsala; BA Gregson, Newcastle upon Tyne; T Howells, Uppsala; K Kiening, Heidelberg; J Mattern, Heidelberg; P Nilsson, Uppsala; I Piper, Glasgow; A Ragauskas, Kaunas; J Sahuquillo, Barcelona; YH Yau, Edinburgh.

BrainIT Group – Data Contributors: Professor Per Enblad, Department of Clinical Neurosciences, Section of Neurosurgery, Uppsala University Hospital, S-75185 Uppsala, Sweden. Phone: +46 18 611 0000 Fax: +46 18 558617 Email:;Dr Karl Kiening, Department of Neurosurgery, Im Neuenheimer Feld 672, 69120 Heidelberg, Germany Phone: +49 30 4505 60724, Fax: +49 30 4505 6072, Email; Dr Giuseppe Citerio, Dipartimento di Anestesia e Rianimazione, Azienda Ospedaliera, Ospedale San Gerardo, Via Donizetti 106, 20052 Monza (Milano), Italy. Phone: +39 029 233 3293, Fax: +39 029 233 3293, Email:; Professor Juan Sahuquillo, Department of Neurosurgery, Neurotraumatology Research Unit, Institut Catala de la Salut, Paseo Vall d’Hebron, 119-129, 08035 Barcelona, Spain. Phone: +34 93 489 3512, Fax: +34 93 489 3513, Email:; Professor Arminas Ragauskas, Telematics Science Laboratory, Kaunas University of Technology, Studentu 50-449, LT3031 Kaunas, Lithuania. Phone: +370 7 736897, Fax: +370 7 736 6869, Email:; Dr Iain Chambers, Regional Medical Physics Department, Newcastle General Hospital, Westgate Road, Newcastle upon Tyne, NE4 6BE, UK. Phone: +44 191 273 8811, Fax: +44 191 226 0970, Email:; Professor Ian Whittle, Department of Clinical Neurosciences, Western General Infirmary, Crewe Road, Edinburgh, UK. Phone: +44 131 537 2103, Fax: +44 131 537 2561, Email:; Professor David Wyper, Department of Clinical Physics, Institute of Neurological Sciences, Southern General Hospital, 1345 Govan Road, Glasgow G51 4TF, UK. Phone: +44 (0) 141 201 2105, Fax: +44 (0) 141 201 4127, Email:; Dr Michael Kiefer, Department of Neurosurgery, Universitatsklinik des Saarlandes, Kirrbergerstrasse, Gebaude 90, D-66421 Homburg-Saar, Germany. Phone: +49 7841 184469, Fax: +49 6841 164480, Email:; Professor Peter Vajkoczy, Department of Neurosurgery, University Hospital Mannheim, Theodor-Kutzer-Ufer 1-3, D68167 Mannheim, Germany. Phone: +49 621 3832360, Fax: +49 621 3832004, Email:; Dr Dirk de Jong, Department of Neurosurgery, University Hospital Rotterdam, Box 2040, Dr Molewaterplein 40, 3015GD Rotterdam, The Netherlands. Phone: +31 10 4087825, Fax: +31 10 4089452, Email:; Professor Flemming Gjerris, University Clinic of Neurosurgery, Rigshospitalet, Blegdamsvej 9, DK 2100 Copenhagen, Denmark. Phone: +45 35 452390, Fax: +45 35 456575, Email:; Professor Nino Stocchetti, Servizio di Anestesia e Rianimazione, Terapia Intensiva Neuroscienze, Via Francesco Sforza 35, 20122 Milano, Italy. Phone: +39 02 550 35517, Fax: +39 02 599 02239, Email:; Dr Luciana Mascia, Department of Anaesthesia and Intensive Care Medicine, Ospedale Molinette, Corso Bramante 88, 10126 Torino, Italy. Email:; Professor Dr Reto Stocker, Division of Surgical Intensive Care, Universitasspital Zurich, B-HOF109, Ramistrasse 100, 8091 Zurich, Switzerland. Phone: +41 1 255 2376, Fax: +41 1 255 3172, Email:; Professor Jan Goffin, Department of Neurosurgery, University Hospital Gasthuisberg, Herestraat 49, 3000 Leuven, Belgium. Phone: +32 16 344290, Fax: +32 16 344285, Email:; Dr Bertil Rydenhag, Institute of Clinical Neuroscience, Department of Neurosurgery, Sahlgrenska University Hospital, Bla Straket 7, van 5, SE 413-45 Gothenburg, Sweden. Phone: +45 31 3421578, Fax: +45 31 416719, Email:; Professor John Pickard, Academic Neurosurgical Unit, Addenbrookes Hospital, Level 4, A Block, Box 167, Hills Road, Cambridge CB2 2QQ, UK. Phone: +44 01223 336946, Fax: +44 01223 3216926, Email:; Dr Sue Hill, Department of Anaesthetics, Southampton General Hospital, Tremona Road, Southampton SO16 6YD, UK. Phone: +44 02380 796401, Fax: +44 02380 794843, Email:; Professor Garth Cruickshank, University Department of Neurosurgery, Queen Elizabeth Hospital, Edgbaston, Birmingham B15 2TH, UK. Phone: +44 0121 697 8225, Fax: +44 0121 697 8248, Email:; Dr Stefan Iencean, Department of Neurosurgery, SF Treime Hospital, 2 Ateneului Street, 6600 Iasi, Romania. Email:; Mr Lawrence Watkins, Department of Neurosurgery, National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, UK. Phone: +44 0207 837 3611, Email:

Compliance with ethical standards

Conflict of interest

We declare that we have no conflicts of interest.


  1. 1.
    American College of Surgeons. Advanced trauma life support. 2011.Google Scholar
  2. 2.
    Jones P, Andrews P, Midgley S, Anderson S, Piper I, Tocher J, Housley A, Corrie J, Slattery J, Dearden M, Douglas Miller J. Measuring the burden of secondary insults in head-injured patients during intensive care. J Neurosurg Anesthesiol. 1994;6:4–14.CrossRefGoogle Scholar
  3. 3.
    Marmarou A, Anderson RL, Ward JD, et al. Impact of ICP instability and hypotension on outcome in patients with severe head trauma. J Neurosurg. 1991;75:S59–66.CrossRefGoogle Scholar
  4. 4.
    BrainIT Consortium. BrainIT. 2007.Google Scholar
  5. 5.
    BrainIT. Brain-it core dataset, manual of operations, data element definitions. 2009.Google Scholar
  6. 6.
    Piper I, Chambers I, Citerio G, Enblad P, Gregson B, Howells T, Kiening K, Mattern J, Nilsson P, Ragauskas A, Sahuquillo J, Donald R, Sinnott R, Stell A. The brain monitoring with information technology (BrainIT) collaborative network: EC feasibility study results and future direction. Acta Neurochir. 2010;152(11):1859–71. Scholar
  7. 7.
    AvertIT. Avert-it project (FP7-217049-AVERT-IT). 2008.Google Scholar
  8. 8.
    Funahashi K. On the approximate realization of continuous mappings by neural networks. Neural Networks. 1989;2(3):183–192.CrossRefGoogle Scholar
  9. 9.
    Bishop CM. Neural networks for pattern recognition. Oxford University Press; 1995.Google Scholar
  10. 10.
    Ripley BR. Pattern recognition and neural networks. Cambridge University Press; 1996.Google Scholar
  11. 11.
    MacKay DJC. Probable networks and plausible predictions—a review of practical bayesian methods for supervised neural networks. Network. 1995;6:469–505.CrossRefGoogle Scholar
  12. 12.
    Neal RM. Bayesian learning for neural networks. Springer, Lecture Notes in Statistics. 1996.Google Scholar
  13. 13.
    Stell A, Sinnott R, Jiang J, Donald R, Chambers I, Citerio G, Enblad P, Gregson B, Howells T, Kiening K, Nilsson P, Ragauskas A, Sahuquillo J, Piper I. Federating distributed clinical data for the prediction of adverse hypotensive events. Philos Trans R Soc. 2009;367(1898):2679–90.CrossRefGoogle Scholar
  14. 14.
    Hanfelt JJ, Slack RS, Gehan EA. A modification of simon’s optimal design for phase ii trials when the criterion is median sample size. Control Clin Trials. 1999;20:555–66.CrossRefGoogle Scholar
  15. 15.
    Simon R. Optimal two-stage designs for phase ii clinical trials. Control Clin Trials. 10:1–10, 1989.CrossRefGoogle Scholar
  16. 16.
    Crawley M. The R book (Second Edition). New York: Wiley; 2012.Google Scholar
  17. 17.
    Kumar A, Roberts D, Wood KE, Light B, Parrillo JE, Sharma S, Suppes R, Feinstein D, Zanotti S, Taiberg L, Gurka D, Kumar A, Cheang M. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med. 2006;34(6):1589–96.CrossRefGoogle Scholar
  18. 18.
    Bijker JB, vanKlei WA, Vergouwe Y, .Eleveld DJ, van Wolfswinkel L, Moons KGM, Kalkman CJ. Intraoperative hypotension and 1-year mortality after noncardiac surgery. Anesthesiology. 2009;111(6):1217–26.CrossRefGoogle Scholar
  19. 19.
    Smiley R. Fast fourier transforms as prophecy: predicting hypotension during spinal anes-thesia. Anesthesiology. 2005;102(6):1079–80.CrossRefGoogle Scholar
  20. 20.
    Moody GB, Lehman LH. Predicting acute hypotensive episodes: the10th annual physionet/computers in cardiology challenge. Comput Cardiol. 2009;36:541–4.Google Scholar
  21. 21.
    Henriques JH, Rocha TR. Prediction of acute hypotensive episodes using neural network multi-models. Comput Cardiol. 2009;36:549–52.Google Scholar
  22. 22.
    Chen X, Xu D, Zhang G, Mukkamala R. Forecasting acute hypotensive episodes in intensive care patients based on a peripheral arterial blood pressure waveform. Comput Cardiol. 2009;36:545–8.Google Scholar
  23. 23.
    Mneimneh MA, Povinelli RJ. A rule-based approach toward the prediction of acute hypotensive episodes. Comput Cardiol. 2009;36:557–60.Google Scholar
  24. 24.
    Gabutti L, Vadilonga D, Mombelli G, Burnier M, Marone C. Artificial neural networks improve the prediction of kt/v, follow-up dietary protein intake and hypotension risk in haemodialysis patients. Nephrol Dial Transplant. 2004;19:1204–11.CrossRefGoogle Scholar
  25. 25.
    Reich DL, Hossain S, Krol M, Baez B, Patel P, Bernstein A, Bodian CA. Predictors of hypotension after induction of general anesthesia. Anesth Analg. 2005;101(3):622–8.CrossRefGoogle Scholar
  26. 26.
    Lehman L, Saeed M, Moody G, Mark R. Similarity-based searching in multi-parameter time series databases. Comput Cardiol. 2008;35:653–6.Google Scholar
  27. 27.
    Dasta JF, McLaughlin TP, Mody SH, Piech CT. Daily cost of an intensive care unit day: the contribution of mechanical ventilation. Crit Care Med. 2005;33(6):1266–71.CrossRefGoogle Scholar
  28. 28.
    Donald R, Howells T, Piper I, Chambers I, Citerio G, Enblad P, Gregson B, Kiening K, Mattern J, Nilsson P, Ragauskas A, Sahuquillo J, Sinnott R, Stell A. Trigger characteristics of EUSIG-defined hypotensive events. Acta Neurochir Suppl. 2012;114:45 – 9.CrossRefGoogle Scholar
  29. 29.
    Williams CKI, Georgatzis K, Hawthorne C, McMonagle P, Piper I, Shaw M, Lal P. Detecting artifactual events in vital signs monitoring data. In: Clifton DA, editor. Machine learning for healthcare technologies; 2016.
  30. 30.
    Ghassemi M, Celi L,A, Stone J. State of the art review: the data revolution in critical care. Crit Care. 2015;19:118.CrossRefGoogle Scholar
  31. 31.
    IBM InfoSphere, Big Data Help Toronto Hospital Monitor Premature Infants.
  32. 32.
    Docherty AB, Lone NI. Exploiting big data for critical care research. Curr Opin Crit Care. 2015;21(5):467–72.CrossRefGoogle Scholar
  33. 33.
    Celi LA, Mark RG, Stone DJ, Montgomery RA. “Big Data” in the intensive care unit. Closing the data loop. Am J Respir Crit Care Med. 2013;187(11):1157–60.CrossRefGoogle Scholar
  34. 34.
    Figini S, Maggi M. Performance of credit risk prediction models via proper loss functions. Technical report, Universita di Pavia, Department of Economics and Management, 2014.Google Scholar
  35. 35.
    Hand D. Assessing the performance of classification methods. Int Stat Rev. 2012;80(3):400–14. Scholar

Copyright information

© Springer Science+Business Media B.V., part of Springer Nature 2018

Authors and Affiliations

  • Rob Donald
    • 1
  • Tim Howells
    • 2
  • Ian Piper
    • 3
    Email author return OK on get
  • P. Enblad
    • 4
  • P. Nilsson
    • 4
  • I. Chambers
    • 5
  • B. Gregson
    • 6
  • G. Citerio
    • 7
  • K. Kiening
    • 8
  • J. Neumann
    • 8
  • A. Ragauskas
    • 9
  • J. Sahuquillo
    • 10
  • R. Sinnott
    • 11
  • A. Stell
    • 12
  • the BrainIT Group
  1. 1.Stats Research LtdDingwallScotland, UK
  2. 2.Department of NeurosurgeryUppsala University HospitalUppsalaSweden
  3. 3.Institute of Neurological SciencesQueen Elizabeth University Hospital, Clinical PhysicsGlasgowScotland, UK
  4. 4.Section of NeurosurgeryUppsala UniversityUppsalaSweden
  5. 5.Department of Medical PhysicsJames Cook University HospitalMiddlesbroughUK
  6. 6.Neurosurgical Trials GroupNewcastle UniversityNewcastle upon TyneUK
  7. 7.Hospital San Gerardo, NeurorianimazioneMonzaItaly
  8. 8.Department of NeurosurgeryRuprecht-Karls-Universitat HospitalHeidelbergGermany
  9. 9.Kaunas University of TechnologyKaunasLithuania
  10. 10.Department of NeurosurgeryVall d’Hebron University HospitalBarcelonaSpain
  11. 11.Department of Information SystemsUniversity of MelbourneParkvilleAustralia
  12. 12.Department of Clinical PhysicsUniversity of GlasgowGlasgowScotland, UK

Personalised recommendations