Journal Home > Volume 28 , Issue 4

The mortality rate in the intensive care unit (ICU) is a key metric of hospital clinical quality. To enhance hospital performance, many methods have been proposed for the stratification of patients’ different risk categories, such as severity scoring systems and machine learning models. However, these methods make capturing time sequence information difficult, posing challenges to the continuous assessment of a patient’s severity during their hospital stay. Therefore, we built a predictive model that can make predictions throughout the patient’s stay and obtain the patient’s risk of death in real time. Our proposed model performed much better than other machine learning methods, including logistic regression, random forest, and XGBoost, in a full set of performance evaluation processes. Thus, the proposed model can support physicians’ decisions by allowing them to pay more attention to high-risk patients and anticipate potential complications to reduce ICU mortality.


menu
Abstract
Full text
Outline
About this article

Use of Deep Learning for Continuous Prediction of Mortality for All Admissions in Intensive Care Units

Show Author's information Guangjian Zeng1Jinhu Zhuang1Haofan Huang1Mu Tian1Yi Gao1Yong Liu2( )Xiaxia Yu1( )
School of Biomedical Engineering, Health Science Center, Shenzhen University, Shenzhen 518060, China
Shenzhen Hospital, Southern Medical University, Shenzhen 518060, China

Abstract

The mortality rate in the intensive care unit (ICU) is a key metric of hospital clinical quality. To enhance hospital performance, many methods have been proposed for the stratification of patients’ different risk categories, such as severity scoring systems and machine learning models. However, these methods make capturing time sequence information difficult, posing challenges to the continuous assessment of a patient’s severity during their hospital stay. Therefore, we built a predictive model that can make predictions throughout the patient’s stay and obtain the patient’s risk of death in real time. Our proposed model performed much better than other machine learning methods, including logistic regression, random forest, and XGBoost, in a full set of performance evaluation processes. Thus, the proposed model can support physicians’ decisions by allowing them to pay more attention to high-risk patients and anticipate potential complications to reduce ICU mortality.

Keywords:

deep learning, representation learning, mortality, risk prediction, critical care
Received: 09 January 2022 Revised: 12 May 2022 Accepted: 05 July 2022 Published: 06 January 2023 Issue date: August 2023
References(45)
[1]
G. Rouleau, M. P. Gagnon, and J. Côté, Impacts of information and communication technologies on nursing care: An overview of systematic reviews (protocol), Syst. Rev., vol. 4, no. 1, p. 75, 2015.
[2]
K. G. M. Moons, R. F. Wolff, R. D. Riley, P. F. Whiting, M. Westwood, G. S. Collins, J. B. Reitsma, J. Kleijnen, and S. Mallett, PROBAST: A tool to assess risk of bias and applicability of prediction model studies: Explanation and elaboration, Ann. Intern. Med., vol. 170, no.1, pp. W1–W33, 2019.
[3]
R. Lilford and P. Pronovost, Using hospital mortality rates to judge hospital performance: A bad idea that just won’t go away, BMJ, vol. 340, p. c2016, 2010.
[4]
K. M. D. M. Karunarathna, Predicting ICU death with summarized patient data, presented at the 2018 IEEE 8th Annu. Computing and Communication Workshop and Conf. (CCWC), Las Vegas, NV, USA, 2018, pp. 238–247.
[5]
R. T. Thomson, D. Leuttel, F. Healey, and S. Scobie, Safer Care for the Acutely Ill Patient: Learning from Serious Incidents. London, UK: National Patient Safety Agency, 2007.
[6]
R. Sadeghi, T. Banerjee, and W. Romine, Early hospital mortality prediction using vital signals, Smart Health, vols. 9–10, pp. 265–274, 2018.
[7]
B. E. Keuning, T. Kaufmann, R. Wiersema, A. Granholm, V. Pettilä, M. H. Møller, C. F. Christiansen, J. C. Forte, H. Snieder, F. Keus, et al., Mortality prediction models in the adult critically ill: A scoping review, Acta Anaesthesiol. Scand., vol. 64, no. 4, pp. 424–442, 2020.
[8]
S. Lemeshow and J. R. Le Gall, Modeling the severity of illness of ICU patients: A systems update, JAMA, vol. 272, no. 13, pp. 1049–1055, 1994.
[9]
J. I. F. Salluh and M. Soares, ICU severity of illness scores: APACHE, SAPS and MPM, Curr. Opin. Crit. Care, vol. 20, no. 5, pp. 557–565, 2014.
[10]
J. L. Vincent, R. Moreno, J. Takala, S. Willatts, A. De Mendonça, H. Bruining, C. K. Reinhart, P. M. Suter, and L. G. Thijs, The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure, Intensive Care Med., vol. 22, no. 7, pp. 707–710, 1996.
[11]
J. E. Zimmerman, A. A. Kramer, D. S. Mcnair, and F. M. Malila, Acute Physiology and Chronic Health Evaluation (APACHE) IV: Hospital mortality assessment for today’s critically ill patients, Crit. Care Med., vol. 34, no. 5, pp. 1297–1310, 2006.
[12]
W. A. Knaus, E. A. Draper, D. P. Wagner, and J. E. Zimmerman, APACHE II: A severity of disease classification system, Crit. Care Med., vol. 13, no. 10, pp. 818–829, 1985.
[13]
T. L. Higgins, D. Teres, W. Copes, B. Nathanson, M. Stark, and A. Kramer, Updated mortality probability model (MPM-III), Chest, vol. 128, no. 4, p. 348S, 2005.
[14]
K. Strand and H. Flaatten, Severity scoring in the ICU: A review, Acta Anaesthesiol. Scand., vol. 52, no. 4, pp. 467–478, 2008.
[15]
S. Vairavan, L. Eshelman, S. Haider, A. Flower, and A. Seiver, Prediction of mortality in an intensive care unit using logistic regression and a hidden Markov model, presented at the 2012 Computing in Cardiology, Krakow, Poland, 2012, pp. 393–396.
[16]
S. Barnes, E. Hamrock, M. Toerper, S. Siddiqui, and S. Levin, Real-time prediction of inpatient length of stay for discharge prioritization, J. Am. Med. Inform. Assoc., vol. 23, no. e1, pp. e2–e10, 2016.
[17]
N. El-Rashidy, S. El-Sappagh, T. Abuhmed, S. Abdelrazek, and H. M. El-Bakry, Intensive care unit mortality prediction: An improved patient-specific stacking ensemble model, IEEE Access, vol. 8, pp. 133541–133564, 2020.
[18]
G. Gutierrez, Artificial intelligence in the intensive care unit, Crit. Care, vol. 24, no. 1, p. 101, 2020.
[19]
S. Leteurtre, A. Duhamel, B. Grandbastien, F. Proulx, J. Cotting, R. Gottesman, A. Joffe, B. Wagner, P. Hubert, A. Martinot, et al., Daily estimation of the severity of multiple organ dysfunction syndrome in critically ill children, CMAJ, vol. 182, no. 11, pp. 1181–1187, 2010.
[20]
S. Leteurtre, A. Duhamel, V. Deken, J. Lacroix, F. Leclerc, and Groupe Francophone de Réanimation et Urgences Pédiatriques, Daily estimation of the severity of organ dysfunctions in critically ill children by using the PELOD-2 score, Crit. Care, vol. 19, no. 1, p. 324, 2015.
[21]
O. Badawi, X. G. Liu, E. Hassan, P. J. Amelung, and S. Swami, Evaluation of ICU risk models adapted for use as continuous markers of severity of illness throughout the ICU stay, Crit. Care Med., vol. 46, no. 3, pp. 361–367, 2018.
[22]
M. J. Rothman, S. I. Rothman, and J. Beals IV, Development and validation of a continuous measure of patient condition using the electronic medical record, J. Biomed. Inform., vol. 46, no. 5, pp. 837–848, 2013.
[23]
M. J. Rothman, .J. J. Tepas, A. J. Nowalk, J. E. Levin, J. M. Rimar, A. Marchetti, and A. L. Hsiao, Development and validation of a continuously age-adjusted measure of patient condition for hospitalized children using the electronic medical record, J. Biomed. Inform., vol. 66, pp. 180–193, 2017.
[24]
C. W. Hug and P. Szolovits, ICU acuity: Real-time models versus daily models, presented at the AMIA 2009 Symp. Proc., American Medical Informatics Association, San Francisco, CA, USA, 2009, pp. 260–264.
[25]
B. Shickel, T. J. Loftus, L. Adhikari, T. Ozrazgat-Baslanti, A. Bihorac, and P. Rashidi, DeepSOFA: A continuous acuity score for critically ill patients using clinically interpretable deep learning, Sci. Rep., vol. 9, no. 1, p. 1879, 2019.
[26]
G. Clermont, D. C. Angus, S. M. Dirusso, M. Griffin, and W. T. Linde-Zwirble, Predicting hospital mortality for patients in the intensive care unit: A comparison of artificial neural networks with logistic regression models, Crit. Care Med., vol. 29, no. 2, pp. 291–296, 2001.
[27]
G. Meyfroidt, F. Güiza, J. Ramon, and M. Bruynooghe, Machine learning techniques to examine large patient databases, Best Pract. Res. Clin. Anaesthesiol., vol. 23, no. 1, pp. 127–143, 2009.
[28]
S. Kim, W. Kim, and R. W. Park, A comparison of intensive care unit mortality prediction models through the use of data mining techniques, Healthc. Inform. Res., vol. 17, no. 4, pp. 232–243, 2011.
[29]
M. Babaie, S. Kalra, A. Sriram, C. Mitcheltree, S. J. Zhu, A. Khatami, S. Rahnamayan, and H. R. Tizhoosh, Classification and retrieval of digital pathology scans: A new dataset, presented at the 2017 IEEE Conf. Computer Vision and Pattern Recognition Workshops, Honolulu, HI, USA, 2017, pp. 760–768.
[30]
A. Khatami, M. Babaie, A. Khosravi, H. R. Tizhoosh, S. M. Salaken, and S. Nahavandi, A deep-structural medical image classification for a Radon-based image retrieval, presented at the 2017 IEEE 30th Canadian Conf. Electrical and Computer Engineering (CCECE), Windsor, Canada, 2017, pp. 1–4.
[31]
A. Khatami, A. Khosravi, C. P. Lim, and S. Nahavandi, A wavelet deep belief network-based classifier for medical images, in Proc. 23rd Int. Conf. Neural Information Processing, Kyoto, Japan, 2016, pp. 467–474.
[32]
A. Khatami, A. Khosravi, T. Nguyen, C. P. Lim, and S. Nahavandi, Medical image analysis using wavelet transform and deep belief networks, Expert Syst. Appl., vol. 86, pp. 190–198, 2017.
[33]
S. Hochreiter and J. Schmidhuber, Long short-term memory, Neural Comput., vol. 9, no. 8, pp. 1735–1780, 1997.
[34]
Y. LeCun, Y. Bengio, and G. Hinton, Deep learning, Nature, vol. 521, no. 7553, pp. 436–444, 2015.
[35]
E. Laksana, M. Aczon, L. Ho, C. Carlin, D. Ledbetter, and R. Wetzel, The impact of extraneous features on the performance of recurrent neural network models in clinical tasks, J. Biomed. Inform., vol. 102, p. 103351, 2020.
[36]
C. S. Carlin, L. V. Ho, D. R. Ledbetter, M. D. Aczon, and R. C. Wetzel, Predicting individual physiologically acceptable states at discharge from a pediatric intensive care unit, J. Am. Med. Inform. Assoc., vol. 25, no. 12, pp. 1600–1607, 2018.
[37]
M. C. Winter, T. E. Day, D. R. Ledbetter, M. D. Aczon, C. J. L. Newth, R. C. Wetzel, and P. A. Ross, Machine learning to predict cardiac death within 1 hour after terminal extubation, Pediatr. Crit. Care Med., vol. 22, no. 2, pp. 161–171, 2021.
[38]
E. Choi, A. Schuetz, W. F. Stewart, and J. M. Sun, Using recurrent neural network models for early detection of heart failure onset, J. Am. Med. Inform. Assoc., vol. 24, no. 2, pp. 361–370, 2017.
[39]
A. Rajkomar, E. Oren, K. Chen, A. M. Dai, N. Hajaj, M. Hardt, P. J. Liu, X. B. Liu, J. Marcus, M. M. Sun, et al., Scalable and accurate deep learning with electronic health records, NPJ Digit. Med., vol. 1, p. 18, 2018.
[40]
M. Saqib, Y. Sha, and M. D. Wang, Early prediction of sepsis in EMR records using traditional ML techniques and deep learning LSTM networks, presented at the 2018 40th Annu. Int. Conf. IEEE Engineering in Medicine and Biology Society (EMBC), Honolulu, HI, USA, 2018, pp. 4038–4041.
[41]
S. Kannan, G. Yengera, D. Mutter, J. Marescaux, and N. Padoy, Future-state predicting LSTM for early surgery type recognition, IEEE Trans. Med. Imaging, vol. 39, no. 3, pp. 556–566, 2020.
[42]
N. Tomašev, N. Harris, S. Baur, A. Mottram, X. Glorot, J. W. Rae, M. Zielinski, H. Askham, A. Saraiva, V. Magliulo, et al., Use of deep learning to develop continuous-risk models for adverse event prediction from electronic health records, Nat. Protoc., vol. 16, no. 6, pp. 2765–2787, 2021.
[43]
A. E. W. Johnson, T. J. Pollard, L. Shen, L. W. H. Lehman, M. L. Feng, M. Ghassemi, B. Moody, P. Szolovits, L. A. Celi, and R. G. Mark, MIMIC-III, a freely accessible critical care database, Sci. Data, vol. 3, p. 160035, 2016.
[44]
A. J. Vickers, A. M. Cronin, E. B. Elkin, and M. Gonen, Extensions to decision curve analysis, a novel method for evaluating diagnostic tests, prediction models and molecular markers, BMC Med. Inform. Decis. Mak., vol. 8, no. 1, p. 53, 2008.
[45]
K. F. Kerr, M. D. Brown, K. H. Zhu, and H. Janes, Assessing the clinical impact of risk prediction models with decision curves: Guidance for correct interpretation and appropriate use, J. Clin. Oncol., vol. 34, no. 21, pp. 2534–2540, 2016.
Publication history
Copyright
Acknowledgements
Rights and permissions

Publication history

Received: 09 January 2022
Revised: 12 May 2022
Accepted: 05 July 2022
Published: 06 January 2023
Issue date: August 2023

Copyright

© The author(s) 2023.

Acknowledgements

This research was supported by the Key Discipline Fund of Shenzhen Hospital of Southern Medical University (No. 2021-2023ICU), the New-Generation Information Technology by the Scientific Research Platform of Institutions of Higher Education of the Education Department of Guangdong Province (No. 2021ZDZX1014), and the Shenzhen University (SZU) Top Ranking Project (No. 86000000210). We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript.

Rights and permissions

The articles published in this open access journal are distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/).

Return