Abstract
Sepsis is a life-threatening disease that is associated with organ dysfunction. It occurs due to the body’s dysregulated response to infection. It is difficult to identify sepsis in its early stages, this delay in identification has a dramatic effect on mortality rate. Developing prognostic tools for sepsis prediction has been the focus of various studies over previous decades. However, most of these studies relied on tracking a limited number of features, as such, these approaches may not predict sepsis sufficiently accurately in many cases. Therefore, in this study, we concentrate on building a more accurate and medically relevant predictive model for identifying sepsis. First, both NSGA-II (a multi-objective genetic algorithm optimization approach) and artificial neural networks are used concurrently to extract the optimal feature subset from patient data. In the next stage, a deep learning model is built based on the selected optimal feature set. The proposed model has two layers. The first is a deep learning classification model used to predict sepsis. This is a stacking ensemble of neural network models that predicts which patients will develop sepsis. For patients who were predicted to have sepsis, data from their first six hours after admission to the ICU are retrieved, this data is then used for further model optimization. Optimization based on this small, recent timeframe leads to an increase in the effectiveness of our classification model compared to other models from previous works. In the second layer of our model, a multitask regression deep learning model is used to identify the onset time of sepsis and the blood pressure at that time in patients that were predicted to have sepsis by the first layer. Our study was performed using the medical information from the intensive care MIMIC III real-world dataset. The proposed classification model achieved 0.913, 0.921, 0.832, 0.906 for accuracy, specificity, sensitivity, and AUC, respectively. In addition, the multitask regression model obtained an RMSE of 10.26 and 9.22 for predicting the onset time of sepsis and the blood pressure at that time, respectively. There are no other studies in the literature that can accurately predict the status of sepsis in terms of its onset time and predict medically verifiable quantities like blood pressure to build confidence in the onset time prediction. The proposed model is medically intuitive and achieves superior performance when compared to all other current state-of-the-art approaches.
Similar content being viewed by others
References
Gaieski DF, Edwards JM, Kallan MJ, Carr BG (2013) Benchmarking the incidence and mortality of severe sepsis in the United States. Crit Care Med 41(5):1167–1174
Fein AM, Balk RA, Knaus WA, Schein RMH, Dellinger RP and Sibbald W (1991) Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis
Levy MM et al. (2003) International sepsis definitions conference, pp. 530–538
Challenges I and Standard G (2015) The third international consensus definitions for sepsis and septic shock ( Sepsis-3 ), 18(2):162–164
Outcomes M (2015) Mortality related to severe sepsis and septic shock among critically Ill patients in Australia and New Zealand, 2000–2012
Daviaud F et al. (2015) Timing and causes of death in septic shock. Ann Intens Care
Layeghian Javan S, Sepehri MM, Layeghian Javan M, Khatibi T (2019) An intelligent warning model for early prediction of cardiac arrest in sepsis patients. Comput Methods Progr Biomed 178:47–58
He Z, Du L, Zhang P, Zhao R, Chen X, and Fang Z (2020) Early sepsis prediction using ensemble learning with deep features and artificial features extracted from clinical electronic health records. pp. 1337–1342
Opal SM, Rubenfeld GD, Van Der Poll T, Vincent J, and Angus DC (2016) The third international consensus definitions for sepsis and septic shock (Sepsis-3), 315(8):801–810
Rhodes A et al. (2017) Surviving sepsis campaign: international guidelines for management of sepsis and septic shock, 2016, 43(3). Springer: Berlin Heidelberg
Calvert JS et al (2016) A computational approach to early sepsis detection. Comput Biol Med 74:69–73
Mao Q et al (2018) Multicentre validation of a sepsis prediction algorithm using only vital sign data in the emergency department, general ward and ICU. BMJ Open 8(1):1–11
Desautels T et al. (2016) Prediction of sepsis in the intensive care unit with minimal electronic health record data: a machine learning approach corresponding author 4:1–15
Verdonk F, Blet A, and Mebazaa A (2017) The new sepsis definition: limitations and contribution to research and diagnosis of sepsis. Curr Opin Anesthesiol, 30(2)
Buehler SS et al (2016) Effectiveness of practices to increase timeliness of providing targeted therapy for inpatients with bloodstream infections: a laboratory medicine best practices systematic review and meta-analysis. Clin Microbiol Rev 29(1):59–103
El-Sappagh S, Abuhmed T, Riazul Islam SM, and Kwak KS (2020) Multimodal multitask deep learning model for alzheimer’s disease progression detection based on time series data. Neurocomputing
Abuhmed T, El-sappagh S, Alonso JM (2021) Robust hybrid deep learning models for Alzheimer’s progression detection. Knowl Based Syst 213:106688
Si Y and Roberts K, Deep patient representation of clinical notes via multi-task learning for mortality prediction, pp. 1–10
Garg A, Mago V (2021) Role of machine learning in medical research: a survey. Comput Sci Rev 40:100370
Kam HJ, Kim HY (2017) Learning representations for the early detection of sepsis with deep neural networks. Comput Biol Med 89:248–255
Chen C-Y, and Huang J-J (2020) Integration of genetic algorithms and neural networks for the formation of the classifier of the hierarchical choquet integral. Inf Sci (Ny)
Fang H, Wang Q, Tu YC, Horstemeyer MF (2008) An efficient non-dominated sorting method for evolutionary algorithms. Evol Comput 16(3):355–384
Hamdani TM, Won JM, Alimi AM, and Karray F (2007) Multi-objective feature selection with NSGA II, Lect Notes Comput Sci (including Subser Lect Notes Artif Intell Lect Notes Bioinformatics), 4431 LNCS, no. Part 1, pp. 240–247
Yusoff Y, Ngadiman MS, Zain AM (2011) Overview of NSGA-II for optimizing machining process parameters. Procedia Eng 15:3978–3983
De Silva N, Ranasinghe M, De Silva CR, and Thurairajah N (2012) Architecture of ensemble neural networks for risk analysis. In: 48th ASC Ann Int Conf Proc, no. April, pp. 1–9
Hansen LK, Salamon P (1990) Neural network ensembles. IEEE Trans Pattern Anal Mach Intell 12(10):993–1001
Perrone MP and Cooper LN (1995) When networks disagree: Ensemble methods for hybrid neural networks, no. February, pp. 342–358
Maclin R (2016) Popular ensemble methods: an empirical study popular ensemble methods: an empirical study, 11:169–198
Shashikumar SP et al (2017) Early sepsis detection in critical care patients using multiscale blood pressure and heart rate dynamics. J Electrocardiol 50(6):739–743
Hug ATRCW, Clifford GD (2012) Clinician blood pressure documentation of stable intensive care patients: an intelligent archiving agent has a higher association with future hypotension, 39(5):1006–1014
Lehman LH, Mark RG, and Nemati S (2016) A model-based machine learning approach to probing autonomic regulation from nonstationary vital-sign time series, 2194(c):1–11
Singer M et al (2016) The third international consensus definitions for sepsis and septic shock (sepsis-3). JAMA 315(8):801–810
Kaukonen K-M, Bailey M, Bellomo R (2015) Systemic inflammatory response syndrome criteria for severe sepsis. The New Engl J Med 373(9):881
Churpek MM, Zadravecz FJ, Winslow C, Howell MD, Edelson DP (2015) Incidence and prognostic value of the systemic inflammatory response syndrome and organ dysfunctions in ward patients. Am J Respir Crit Care Med 192(8):958–964
Whippy A et al (2011) Kaiser Permanente’s performance improvement system, part 3: multisite improvements in care for patients with sepsis. Jt Comm J Qual patient Saf 37(11):483–493
Shankar-Hari M et al (2016) Developing a newdefinition and assessing newclinical criteria for Septic shock: for the third international consensus definitions for sepsis and septic shock (sepsis-3). JAMA 315(8):775–787
Seymour CW et al (2016) Assessment of clinical criteria for sepsis for the third international consensus definitions for sepsis and septic shock (sepsis-3). JAMA 315(8):762–774
Marik PE, Taeb AM (2017) SIRS, qSOFA and new sepsis definition. J Thorac Dis 9(4):943–945
Schinkel M, Paranjape K, Nannan Panday RS, Skyttberg N, Nanayakkara PWB (2019) Clinical applications of artificial intelligence in sepsis: a narrative review. Comput Biol Med 115:103488
Darwiche A, Mukherjee S (2018) Machine learning methods for septic shock prediction. ACM Int Conf Proc Ser 1051:104–110
Arabi Y, Al Shirawi N, Memish Z, Venkatesh S, Al-Shimemeri A (2003) Assessment of six mortality prediction models in patients admitted with severe sepsis and septic shock to the intensive care unit: a prospective cohort study. Crit Care 7(5):R116–R122
Liu R et al (2019) Data-driven discovery of a novel sepsis pre-shock state predicts impending septic shock in the ICU. Sci Rep 9(1):1–9
Ghosh S, Li J, Cao L, Ramamohanarao K (2017) Septic shock prediction for ICU patients via coupled HMM walking on sequential contrast patterns. J Biomed Inform 66:19–31
Scherpf M, Gräßer F, Malberg H, Zaunseder S (2019) Predicting sepsis with a recurrent neural network using the MIMIC III database. Comput Biol Med 113:103395
Fagerström J, Bång M, Wilhelms D, Chew MS (2019) LiSep LSTM: a machine learning algorithm for early detection of septic shock. Sci Rep 9(1):1–8
Song W, Jung SY, Baek H, Choi CW, Jung YH, Yoo S (2020) A predictive model based on machine learning for the early detection of late-onset neonatal sepsis: development and observational study. JMIR Med Inform 8(7):e15965
Yao R et al (2020) A machine learning-based prediction of hospital mortality in patients with postoperative sepsis. Front Med 7:445
Jones CN et al (2014) Spontaneous neutrophil migration patterns during sepsis after major burns. PLoS ONE 9(1):e114509
Lukaszewski RA et al. (2008) Presymptomatic prediction of sepsis in intensive care unit patients, 15(7):1089–1094
Mitra A and Ashraf K (2018) Sepsis prediction and vital signs ranking in intensive care unit patients
Calvert J et al (2017) Cost and mortality impact of an algorithm-driven sepsis prediction system. J Med Econ 20(6):646–651
Desautels T et al (2016) Prediction of sepsis in the intensive care unit with minimal electronic health record data: a machine learning approach. JMIR Med Inform 4(3):e28
Johnson LSAEW, Pollard TJ (2016) Data Descriptor: MIMIC-III, a freely accessible critical care database. Thromb Haemost 76(2):258–262
Scherpf M, Zaunseder S, and Dortmund A (2019) Predicting sepsis with a recurrent neural network using the MIMIC III database Predicting sepsis with a recurrent neural network using the MIMIC III Database, no. August
Stanski NL, Wong HR (2020) Prognostic and predictive enrichment in sepsis. Nat Rev Nephrol 16(1):20–31
Pettinati MJ, Chen G, Rajput KS, and Selvaraj N (2020) Practical machine learning-based sepsis prediction. In: 2020 42nd annual international conference of the IEEE engineering in medicine biology society (EMBC), pp. 4986–4991
Chen M, Hernández A (2021) Towards an explainable model for sepsis detection based on sensitivity analysis. IRBM 1:1–12
Schellenberger S, Shi K, Wiedemann JP, Lurz F, and Weigel R, An ensemble LSTM architecture for clinical sepsis detection, pp. 2–5
Zabihi M, Kiranyaz S, and Gabbouj M (2019) Sepsis prediction in intensive care unit using ensemble of XGboost models, in 2019 computing in cardiology (CinC), 1– 4
Zhang D et al (2021) An interpretable deep-learning model for early prediction of sepsis in the emergency department. Patterns 2(2):100196
Rafiei A, Rezaee A, Hajati F, Gheisari S, Golzan M (2021) SSP: early prediction of sepsis using fully connected LSTM-CNN model. Comput Biol Med 128:104110
Tsang G and Xie X (2021) Deep learning based sepsis intervention: the modelling and prediction of severe sepsis onset. In: 2020 25th international conference on pattern recognition (ICPR), pp. 8671–8678
Al-mualemi BY and Lu LU (2021) A deep learning-based sepsis estimation scheme, 9
Goh KH et al (2021) Artificial intelligence in sepsis early prediction and diagnosis using unstructured data in healthcare. Nat Commun 12(1):711
Aşuroğlu T, Oğul H (2021) A deep learning approach for sepsis monitoring via severity score estimation. Comput Methods Prog Biomed 198:105816
Yuan K-C et al (2020) The development an artificial intelligence algorithm for early sepsis diagnosis in the intensive care unit. Int J Med Inform 141:104176
Liu R, Greenstein JL, Fackler JC, Bergmann J, Bembea MM, and Winslow RL (2021) Prediction of impending septic shock in children with sepsis. Crit Care Explor, 3(6)
Ngufor C, Upadhyaya S, Murphree D, Kor D, and Pathak J (2015) Multi-task learning with selective cross-task transfer for predicting bleeding and other important patient outcomes. In: 2015 IEEE International conference on data science and advanced analytics (DSAA), pp. 1–8
Giacobbe DR et al. (2021) Early detection of sepsis with machine learning techniques: a brief clinical perspective, 8
Fleuren LM et al (2020) Machine learning for the prediction of sepsis: a systematic review and meta-analysis of diagnostic test accuracy. Intens Care Med 46(3):383–400
Vellido A, Ribas V, Morales C, Ruiz Sanmartín A, Ruiz Rodríguez JC (2018) Machine learning in critical care: state-of-the-art and a sepsis case study. Biomed Eng Online 17(S1):1–18
Le S et al (2019) Pediatric severe sepsis prediction using machine learning. Front Pediatr 7:1–8
Kausch SL, Moorman JR, Lake DE, Keim-Malpass J (2021) Physiological machine learning models for prediction of sepsis in hospitalized adults: an integrative review. Intens Crit Care Nurs 65:103035
Hassan N et al (2021) Preventing sepsis how can artificial intelligence inform the clinical decision-making process a systematic review. Int J Med Inform 150:104457
Xie Y et al (2021) A prediction model of sepsis - associated acute kidney injury based on antithrombin III. Clin Exp Med 21(1):89–100
Calvert J et al (2016) Using electronic health record collected clinical variables to predict medical intensive care unit mortality. Ann Med Surg 11:52–57
Fullerton JN, Price CL, Silvey NE, Brace SJ, Perkins GD (2012) Is the Modified early warning score ( MEWS ) superior to clinician judgement in detecting critical illness in the pre-hospital environment? Resuscitation 83(5):557–562
Chandrashekar G, Sahin F (2014) A survey on feature selection methods. Comput Electr Eng 40(1):16–28
Bucholc M et al (2019) A practical computerized decision support system for predicting the severity of Alzheimer’s disease of an individual. Exp Syst Appl 130:157–171
Jain D, Singh V (2018) Feature selection and classification systems for chronic disease prediction: a review. Egypt Inform J 19(3):179–189
Kohavi R, John GH (1997) Wrappers for feature subset selection. Artif Intell 97(1–2):273–324
Sharma M and Kaur P (2021) A comprehensive analysis of nature-inspired meta-heuristic techniques for feature selection problem. Arch Comput Methods Eng, 28(3)
Nguyen BH, Xue B, Zhang M (2020) A survey on swarm intelligence approaches to feature selection in data mining. Swarm Evol Comput 54:100663
Vieira SM, Mendonça LF, Farinha GJ, Sousa JMC (2013) Modified binary PSO for feature selection using SVM applied to mortality prediction of septic patients. Appl Soft Comput 13(8):3494–3504
Li Y, Li T, Liu H (2017) Recent advances in feature selection and its applications. Knowl Inf Syst 53(3):551–577
Huang B, Buckley B, Kechadi T-M (2010) Multi-objective feature selection by using NSGA-II for customer churn prediction in telecommunications. Exp Syst Appl 37(5):3638–3646
Kurnaz S, Mohammed MS, and Mohammed SJ (2020) A high efficiency thyroid disorders prediction system with non-dominated sorting genetic algorithm NSGA-II as a feature selection algorithm. In: 2020 International Conference for Emerging Technology (INCET), pp. 1–6 s
Zhang Y, Gong D, Cheng J (2015) Multi-objective particle swarm optimization approach for cost-based feature selection in classification. IEEE/ACM Trans Comput Biol Bioinforma 14(1):64–75
Khan A, Baig AR (2015) Multi-objective feature subset selection using non-dominated sorting genetic algorithm. J Appl Res Technol 13(1):145–159
Zangooei MH, Habibi J, Alizadehsani R (2014) Disease diagnosis with a hybrid method SVR using NSGA-II. Neurocomputing 136:14–29
Soui M, Mansouri N, Alhamad R, Kessentini M and Ghedira K (2021) NSGA-II as feature selection technique and AdaBoost classifier for COVID-19 prediction using patient’s symptoms. Nonlinear Dyn. pp. 1–23
Soyel H, Tekguc U, Demirel H (2011) Application of NSGA-II to feature selection for facial expression recognition. Comput Electr Eng 37(6):1232–1240
Zitzler E, Deb K, Thiele L (2000) Comparison of multiobjective evolutionary algorithms: empirical results. Evol Comput 8(2):173–195
Salmanpour MR et al (2019) Optimized machine learning methods for prediction of cognitive outcome in Parkinson’s disease. Comput Biol Med 111:103347
Türkşen Ö, Vieira SM, Madeira JFA, Apaydin A, and Sousa JMC (2013) Comparison of multi-objective algorithms applied to feature selection,” in Towards advanced data analysis by combining soft computing and statistics, Springer, pp. 359–375
Hojjati A, Monadi M, Faridhosseini A, Mohammadi M (2018) Application and comparison of NSGA-II and MOPSO in multi-objective optimization of water resources systems. J Hydrol Hydromech 66(3):323–329
Wang R (2016) An improved nondominated sorting genetic algorithm for multiobjective problem. Math Probl Eng, 2016
Khan A and Baig A, Multi-objective feature subset sorting genetic algorithm selection using. J Appl Res Technol, 13(1):145–159
Wang Q, Wang L, Huang W, Wang Z, Liu S, and Savić DA (2019) Parameterization of NSGA-II for the optimal design of water distribution systems. Water (Switzerland), 11(5)
Zhang C and Ma X (2015) NSGA-II algorithm with a local search strategy for multiobjective optimal design of dry-type air-core reactor. Math Probl Eng, 2015
Indexed S, Bachri OS, Program AI, Hatta M, and Nurhayati OD (2017) Feature selection based on chi square in artificial neural network to predict the accuracy of student, 8(8):731–739
Nti IK, Adekoya AF, and Weyori BA (2020) A comprehensive evaluation of ensemble learning for stock-market prediction. J Big Data
Zhang Y and Yang Q (2017) A survey on multi-task learning, pp. 1–20
Ruder S (2017) An overview of multi-task learning in deep neural networks
Chen W, Long G, Yao L, and Sheng QZ (2019) AMRNN: attended multi-task recurrent neural networks for dynamic illness severity prediction
Harutyunyan H, Khachatrian H, Kale DC, Ver Steeg G, Galstyan A (2019) Multitask learning and benchmarking with clinical time series data. Sci Data 6(1):96
Razavian N, Marcus J, and Sontag D, Lab Tests, pp. 1–27
Joenssen DW and Bankhofer U (2015) Hot deck methods for imputing missing data hot deck methods for imputing missing data the effects of limiting donor usage, no. July 2012
Rahman MM, Davis DN (2013) Addressing the class imbalance problem in medical datasets. Int J Mach Learn Comput 2013:224–228
Li D-C, Liu C-W, Hu SC (2010) A learning method for the class imbalance problem with medical data sets. Comput Biol Med 40(5):509–518
Moreo A and Esuli A (2016) Distributional random oversampling for imbalanced text classification pp. 805–808
Chawla NV, Bowyer KW, Hall LO, and Kegelmeyer WP (2002) SMOTE: synthetic minority over-sampling technique
Li S (2015) ADASYN: adaptive synthetic sampling approach for imbalanced, no. July 2008
Tahir MA, Kittler J, Mikolajczyk K, and Yan F (2009) A multiple expert approach to the class imbalance problem using inverse random under sampling a multiple expert approach to the class imbalance problem using inverse random, no. May 2014
Moon TK (1996) The expectation-maximization algorithm. IEEE Sig Process Mag 13(6):47–60
Elzeki OM, Alrahmawy MF, and Elmougy S (2019) A new hybrid genetic and information gain algorithm for imputing missing values in cancer genes datasets, no. December, pp. 20–33
Azur MJ, Stuart EA, Frangakis C, Leaf PJ, and Washington DC (2011) Multiple imputation by chained equations: what is it and how does it work?, 20(1):40–49
Hotchkiss RS, Moldawer LL, Opal SM, Reinhart K, Turnbull IR, and Vincent JL (2016) Sepsis and septic shock. Nat Rev Dis Prim, 2
F. Mitchell M Levy, MD, MCCM Sean R Townsend, MD (2019) Early identification of sepsis on the hospital floors
Kaji DA et al (2019) An attention based deep learning model of clinical events in the intensive care unit. PLoS ONE 14(2):1–17
Snoek J, Larochelle H and Adams RP, Practical Bayesian optimization of machine learning algorithms, pp. 1–9
Bergstra J and Bengio Y (2012) Random search for hyper-parameter optimization, 13:281–305
Islam MM, Nasrin T, Walther BA, Wu CC, Yang HC, Li YC (2019) Prediction of sepsis patients using machine learning approach: a meta-analysis. Comput Methods Prog Biomed 170:1–9
Wang RZ, Sun CH, Schroeder PH, Ameko MK, Moore CC, Barnes LE (2018) Predictive models of sepsis in adult ICU patients. Proc - 2018 IEEE Int Conf Healthc Informatics, ICHI pp. 390–391
Demˇ J (2006) Statistical comparisons of classifiers over multiple data sets, 7:1–30
Friedman M (1990) The use of ranks to avoid the assumption of normality implicit in the analysis of variance. J Am Stat Assoc 32(200):675–701
Calvert J et al (2016) High-performance detection and early prediction of septic shock for alcohol-use disorder patients. Ann Med Surg 8:50–55
Shashikumar SP, Josef CS, Sharma A, Nemati S (2021) DeepAISE – an interpretable and recurrent neural survival model for early prediction of sepsis. Artif Intell Med 113:102036
Funding
This research was supported by the MSIT(Ministry of Science and ICT), Korea, under the ICT Creative Consilience Program(IITP-2021-2020-0-01821) supervised by the IITP(Institute for Information & communications Technology Planning & Evaluation), and the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (No. 2021R1A2C1011198).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflicts of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
El-Rashidy, N., Abuhmed, T., Alarabi, L. et al. Sepsis prediction in intensive care unit based on genetic feature optimization and stacked deep ensemble learning. Neural Comput & Applic 34, 3603–3632 (2022). https://doi.org/10.1007/s00521-021-06631-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-021-06631-1