Brazilian Journal of Anesthesiology
Brazilian Journal of Anesthesiology
Original Investigation

Predictive model for difficult laryngoscopy using machine learning: retrospective cohort study

Modelo preditivo para laringoscopia difícil usando aprendizado de máquina: estudo de coorte retrospectivo

Jong Ho Kim, Jun Woo Choi, Young Suk Kwon, Seong Sik Kang

Downloads: 0
Views: 346


Both predictions and predictors of difficult laryngoscopy are controversial. Machine learning is an excellent alternative method for predicting difficult laryngoscopy. This study aimed to develop and validate practical predictive models for difficult laryngoscopy through machine learning.

Variables for the prediction of difficult laryngoscopy included age, Mallampati grade, body mass index, sternomental distance, and neck circumference. Difficult laryngoscopy was defined as grade 3 and 4 by the Cormack-Lehane classification. Pre-anesthesia and anesthesia data of 616 patients who had undergone anesthesia at a single center were included. The dataset was divided into a base training set (n = 492) and a base test set (n = 124), with equal distribution of difficult laryngoscopy. Training data sets were trained with six algorithms (multilayer perceptron, logistic regression, supportive vector machine, random forest, extreme gradient boosting, and light gradient boosting machine), and cross-validated. The model with the highest area under the receiver operating characteristic curve (AUROC) was chosen as the final model, which was validated with the test set.

The results of cross-validation were best using the light gradient boosting machine algorithm with Mallampati score x age and sternomental distance as predictive model parameters. The predicted AUROC for the difficult laryngoscopy class was 0.71 (95% confidence interval, 0.59–0.83; p =  0.014), and the recall (sensitivity) was 0.85.

Predicting difficult laryngoscopy is possible with three parameters. Severe damage resulting from failure to predict difficult laryngoscopy with high recall is small with the reported model. The model’s performance can be further enhanced by additional data training.


Intratracheal intubation;  Laryngoscopes;  Machine learning


Justificativa: Tanto as previsões quanto os preditores de laringoscopia difícil são controversos. O aprendizado de máquina é um excelente método alternativo para prever laringoscopias difíceis. Este estudo teve como objetivo desenvolver e validar modelos preditivos práticos para laringoscopia difícil por meio de aprendizado de máquina. Métodos: As variáveis para predição de laringoscopia difícil incluíram idade, grau de Mallampati, índice de massa corporal, distância esternomentoniana e circunferência do pescoço. A laringoscopia difícil foi definida como grau 3 e 4 pela classificação de Cormack-Lehane. Foram incluídos dados pré-anestésicos e anestésicos de 616 pacientes submetidos à anestesia em um único centro. O conjunto de dados foi dividido em um conjunto base de treinamento (n = 492) e um conjunto base de teste (n = 124), com distribuição igual de laringoscopia difícil. Os conjuntos de dados de treinamento foram treinados com seis algoritmos (perceptron multicamada, regressão logística, máquina de vetor de suporte, floresta aleatória, aumento de gradiente extremo e máquina de aumento leve de gradiente) e validação cruzada. O modelo com maior área sob a curva característica de operação do receptor (AUROC) foi escolhido como modelo final, que foi validado com o conjunto de testes. Resultados: Os resultados da validação cruzada foram melhores usando o algoritmo da máquina de aumento leve do gradiente com grau de Mallampati x idade e distância esternomental como parâmetros do modelo preditivo. O AUROC previsto para a classe de laringoscopia difícil foi de 0,71 (intervalo de confiança de 95%, 0,59–0,83; p = 0,014) e a recordação (sensibilidade) foi de 0,85. Conclusão: A previsão de laringoscopia difícil é possível com três parâmetros. Danos graves resultantes da falha em prever laringoscopia difícil com alta evocação são pequenos com o modelo relatado. O desempenho do modelo pode ser aprimorado ainda mais por treinamento de dados adicional.


Intubação intratraqueal; Laringoscópios; Aprendizado de máquina


J.L. Apfelbaum, C.A. Hagberg, R.A. Caplan, et al.
Practice guidelines for management of the difficult airwayan updated report by the American Society of Anesthesiologists task force on management of the difficult airway
Anesthesiology., 118 (2013), pp. 251-270

S.A. Workeneh, A.H. Gebregzi, Z.A. Denu
Magnitude and Predisposing Factors of Difficult Airway during Induction of General Anaesthesia
Anesthesiol Res Pract., 2017 (2017), Article 5836397

F. Etezadi, A. Ahangari, H. Shokri, et al.
Thyromental height: a new clinical test for prediction of difficult laryngoscopy
Anesth Analg., 117 (2013), pp. 1347-1351

C. Frerk
Predicting difficult intubation
Anaesthesia., 46 (1991), pp. 1005-1008

Z.H. Khan, A. Kashfi, E. Ebrahimkhani
A comparison of the upper lip bite test (a simple new technique) with modified Mallampati classification in predicting difficulty in endotracheal intubation: a prospective blinded study
Anesth Analg., 96 (2003), pp. 595-599

S.R. Mallampati, S.P. Gatt, L.D. Gugino, et al.
A clinical sign to predict difficult tracheal intubation; a prospective study
Can Anaesth Soc J., 32 (1985), pp. 429-434

W. Riad, M.N. Vaez, R. Raveendran, et al.
Neck circumference as a predictor of difficult intubation and difficult mask ventilation in morbidly obese patients: A prospective observational study
Eur J Anaesthesiol., 33 (2016), pp. 244-249

D. Savva
Prediction of difficult tracheal intubation
Br J Anaesth., 73 (1994), pp. 149-153

S. Türkan, Y. Ates, H. Cuhruk, et al.
Should we reevaluate the variables for predicting the difficult airway in anesthesiology?
Anesth Analg., 94 (2002), pp. 1340-1344

O. Langeron, P. Cuvillon, C. Ibanez-Esteve, et al.
Prediction of Difficult Tracheal IntubationTime for a Paradigm Change
Anesthesiology., 117 (2012), pp. 1223-1233

M.A. Moustafa, S. El-Metainy, K. Mahar, et al.
Defining difficult laryngoscopy findings by using multiple parameters: A machine learning approach
Egyptian Journal of Anaesthesia., 33 (2017), pp. 153-158

M. Wilson, D. Spiegelhalter, J. Robertson, et al.
Predicting difficult intubation
Br J Anaesth., 61 (1988), pp. 211-216

J. Arne, P. Descoins, J. Fusciardi, et al.
Preoperative assessment for difficult intubation in general and ENT surgery: predictive value of a clinical multivariate risk index
Br J Anaesth., 80 (1998), pp. 140-146

M. Naguib, T. Malabarey, R.A. AlSatli, et al.
Predictive models for difficult laryngoscopy and intubation. A clinical, radiologic and three-dimensional computer imaging study
Can J Anaesth., 46 (1999), p. 748

N.V. Chawla, K.W. Bowyer, L.O. Hall, et al.
SMOTE: synthetic minority over-sampling technique
J Artif Intell Res., 16 (2002), pp. 321-357

D.L. Wilson
Asymptotic properties of nearest neighbor rules using edited data
IEEE Trans Syst Man Cybern. (1972), pp. 408-421

T. Shiga, Wajima Zi, T. Inoue, et al.
Predicting Difficult Intubation in Apparently Normal PatientsA Meta-analysis of Bedside Screening Test Performance
Anesthesiology., 103 (2005), pp. 429-437

S.M. Yentis
Predicting difficult intubation--worthwhile exercise or pointless ritual?
Anaesthesia., 57 (2002), pp. 105-109

A. Géron
Hands-on machine learning with Scikit-Learn and TensorFlow: concepts, tools, and techniques to build intelligent systems
(1st ed), O’Reilly Media, Inc, Sebastopol (CA) (2017)

The Lancet Respiratory Medicine
Opening the black box of machine learning, 6 (2018), p. 801
Available at: [accessed 11 January 2011]

K. Phillips, M. Moucharite, T. Wong, et al.
Cost burden associated with difficult intubation in the United States
Trends Anaesth Crit Care., 30 (2020), p. e131

F. Adnet, S.W. Borron, S.X. Racine, et al.
The intubation difficulty scale (IDS) proposal and evaluation of a new score characterizing the complexity of endotracheal intubation
Anesthesiology., 87 (1997), pp. 1290-1297

Z. Zhuang, D. Landsittel, S. Benson, et al.
Facial anthropometric differences among gender, ethnicity, and age groups
Ann Occup Hyg., 54 (2010), pp. 391-402

60f083b6a953950fa7049f9c rba Articles
Links & Downloads

Braz J Anesthesiol

Share this page
Page Sections