Progress and Perspective of Artificial Intelligence and Machine Learning of Prediction in Anesthesiology

doi:10.1007/s12204-021-2331-3

Abstract

Abstract: Artificial intelligence (AI) has long been an attractive topic in medicine, especially in light of the rapid developments in digital and information technologies. AI has already provided some breakthroughs in medicine. With the assistance of AI, more precise models have been used for clinical predictions, diagnoses, and decision-making. This review defines the basic concepts of AI and machine learning (ML), and provides a simple introduction to certain frequently used algorithms in AI and ML. In addition, the review discusses the current common applications of AI and ML in the prediction of anesthesia conditions, including those for preoperative predictions of difficult airways, intraoperative predictions of adverse events and anesthetic effects, and postoperative predictions of vomiting and pain. The use of AI in anesthesiology remains in development, even without extensive promotion and clinical application; moreover, it has immense potential to maintain further development in the future. Finally, the limitations and challenges of AI development for anesthesia are also discussed, along with considerations regarding ethics and safety.

Key words: artificial intelligence (AI), machine learning (ML), anesthesiology, prediction models

CLC Number:

R 4
TP 181

XIA Ming (夏明), XU Tianyi (徐天意), JIANG Hong∗ (姜虹). Progress and Perspective of Artificial Intelligence and Machine Learning of Prediction in Anesthesiology[J]. J Shanghai Jiaotong Univ Sci, 2022, 27(1): 112-120.

References 67

[1]	JORDAN M I, MITCHELL T M. Machine learning:Trends, perspectives, and prospects [J]. Science, 2015,349(6245): 255-260.
[2]	IBRAHIMA, PRIMAKOV S, BEUQUEM, et al. Radiomicsfor precision medicine: Current challenges, futureprospects, and the proposal of a new framework[J]. Methods, 2021, 188: 20-29.
[3]	ANEJA S, CHANG E, OMURO A. Applications ofartificial intelligence in neuro-oncology [J]. CurrentOpinion in Neurology, 2019, 32(6): 850-856.
[4]	SCHWYZER M, MARTINI K, BENZ D C, et al. Artificialintelligence for detecting small FDG-positivelung nodules in digital PET/CT: Impact of image reconstructionson diagnostic performance [J]. EuropeanRadiology, 2020, 30(4): 2031-2040.
[5]	LAURITSEN S M, KRISTENSENM, OLSEN M V, etal. Explainable artificial intelligence model to predictacute critical illness from electronic health records [J].Nature Communications, 2020, 11(1): 3852.
[6]	GUNASEKERAN D V, TING D S W, TAN G SW, et al. Artificial intelligence for diabetic retinopathyscreening, prediction and management [J]. CurrentOpinion in Ophthalmology, 2020, 31(5): 357-365.
[7]	LOFTUS T J, TIGHE P J, FILIBERTO A C, et al.Artificial intelligence and surgical decision-making [J].JAMA Surgery, 2020, 155(2): 148-158.
[8]	SHORTLIFFE E H, SEP′ULVEDA M J. Clinical decisionsupport in the era of artificial intelligence [J].JAMA, 2018, 320(21): 2199-2200.
[9]	LEE M S, GRABOWSKI M M, HABBOUB G, etal. The impact of artificial intelligence on quality andsafety [J]. Global Spine Journal, 2020, 10(Sup. 1): 99-103.
[10]	HOGARTY D T, MACKEY D A, HEWITT A W.Current state and future prospects of artificial intelligencein ophthalmology: A review [J]. Clinical & ExperimentalOphthalmology, 2019, 47(1): 128-139.
[11]	WANG S Y, PERSHING S, LEE A Y, et al. Bigdata requirements for artificial intelligence [J]. CurrentOpinion in Ophthalmology, 2020, 31(5): 318-323.
[12]	CONNOR C W. Artificial intelligence and machinelearning in anesthesiology [J]. Anesthesiology, 2019,131(6): 1346-1359.
[13]	RAJKOMAR A, DEAN J, KOHANE I.Machine learningin medicine [J]. The New England Journal ofMedicine, 2019, 380(14): 1347-1358.
[14]	HOWARD J. Artificial intelligence: Implications forthe future of work [J]. American Journal of IndustrialMedicine, 2019, 62(11): 917-926.
[15]	HANDELMAN G S, KOK H K, CHANDRA R V, et al.eDoctor: Machine learning and the future of medicine[J]. Journal of Internal Medicine, 2018, 284(6): 603-619.
[16]	MOORE M M, SLONIMSKY E, LONG A D, et al.Machine learning concepts, concerns and opportunitiesfor a pediatric radiologist [J]. Pediatric Radiology,2019, 49(4): 509-516.
[17]	UDDINS, KHANA, HOSSAINME, et al. Comparingdifferent supervised machine learning algorithms fordisease prediction [J]. BMC Medical Informatics andDecision Making, 2019, 19(1): 281.
[18]	CRUZ J A, WISHART D S. Applications of machinelearning in cancer prediction and prognosis [J]. CancerInformatics, 2007, 2: 59-77.
[19]	ZHAO X, WU Y H, LEE D L, et al. iForest: Interpretingrandom forests via visual analytics [J]. IEEETransactions on Visualization and Computer Graphics,2019, 25(1): 407-416.
[20]	HASHIMOTO D A, WITKOWSKI E, GAO L, et al.Artificial intelligence in anesthesiology: Current techniques,clinical applications, and limitations [J]. Anesthesiology,2020, 132(2): 379-394.
[21]	PERGIALIOTIS V, POULIAKIS A, PARTHENIS C,et al. The utility of artificial neural networks and classificationand regression trees for the prediction of endometrialcancer in postmenopausal women [J]. PublicHealth, 2018, 164: 1-6.
[22]	HINTON G. Deep learning: A technology with thepotential to transform health care [J]. JAMA, 2018,320(11): 1101-1102.
[23]	LECUN Y, BENGIO Y, HINTON G. Deep learning[J]. Nature, 2015, 521(7553): 436-444.
[24]	GREGORY A, STAPELFELDT W H, KHANNA AK, et al. Intraoperative hypotension is associated withadverse clinical outcomes after noncardiac surgery [J].Anesthesia and Analgesia, 2021, 132(6): 1654-1665.
[25]	SMISCHNEY N J, SHAWA D, STAPELFELDTWH,et al. Postoperative hypotension in patients dischargedto the intensive care unit after non-cardiac surgery isassociated with adverse clinical outcomes [J]. CriticalCare (London, England), 2020, 24(1): 682.
[26]	HATIB F, JIAN Z P, BUDDI S, et al. Machinelearningalgorithm to predict hypotension based onhigh-fidelity arterial pressure waveform analysis [J].Anesthesiology, 2018, 129(4): 663-674.
[27]	DAVIES S J, VISTISEN S T, JIAN Z P, et al. Abilityof an arterial waveform analysis-derived hypotensionprediction index to predict future hypotensive events in surgical patients [J]. Anesthesia and Analgesia,2020, 130(2): 352-359.
[28]	WIJNBERGE M, GEERTS B F, HOL L, et al. Effectof a machine learning-derived early warning systemfor intraoperative hypotension vs standard careon depth and duration of intraoperative hypotensionduring elective noncardiac surgery: The HYPE randomizedclinical trial [J]. JAMA, 2020, 323(11): 1052-1060.
[29]	MAHESHWARI K, BUDDI S, JIAN Z P, et al. Performanceof the Hypotension Prediction Index with noninvasivearterial pressure waveforms in non-cardiacsurgical patients [J]. Journal of Clinical Monitoringand Computing, 2021, 35(1): 71-78.
[30]	LIN C S, CHANG C C, CHIU J S, et al. Applicationof an artificial neural network to predict postinductionhypotension during general anesthesia [J]. Medical DecisionMaking, 2011, 31(2): 308-314.
[31]	KENDALE S, KULKARNI P, ROSENBERG A D,et al. Supervised machine-learning predictive analyticsfor prediction of postinduction hypotension [J]. Anesthesiology,2018, 129(4): 675-688.
[32]	KANG A R, LEE J, JUNG W, et al. Development ofa prediction model for hypotension after induction ofanesthesia using machine learning [J]. PLoS One, 2020,15(4): e0231172.
[33]	LUNDBERG S M, NAIR B, VAVILALA M S, etal. Explainable machine-learning predictions for theprevention of hypoxaemia during surgery [J]. NatureBiomedical Engineering, 2018, 2(10): 749-760.
[34]	GENG W, TANG H, SHARMA A, et al. An artificialneural network model for prediction of hypoxemiaduring sedation for gastrointestinal endoscopy [J].The Journal of International Medical Research, 2019,47(5): 2097-2103.
[35]	APFEL C C, KRANKE P, EBERHART L H J, etal. Comparison of predictive models for postoperativenausea and vomiting [J]. British Journal of Anaesthesia,2002, 88(2): 234-240.
[36]	EBERHART L H J, H¨OGEL J, SEELING W, et al.Evaluation of three risk scores to predict postoperativenausea and vomiting [J]. Acta AnaesthesiologicaScandinavica, 2000, 44(4): 480-488.
[37]	TRAEGER M, EBERHART A, GELDNER G, et al.Prediction of postoperative nausea and vomiting usingan artificial neural network [J]. Der Anaesthesist, 2003,52(12): 1132-1138.
[38]	PENG S Y, WU K C, WANG J J, et al. Predictingpostoperative nausea and vomiting with the applicationof an artificial neural network [J]. British Journalof Anaesthesia, 2007, 98(1): 60-65.
[39]	GONG C S A, YU L, TING C K, et al. Predictingpostoperative vomiting for orthopedic patients receivingpatient-controlled epidural analgesia with the applicationof an artificial neural network [J]. BioMedResearch International, 2014, 2014: 786418.
[40]	WU H Y, GONG C A, LIN S P, et al. Predicting postoperativevomiting among orthopedic patients receivingpatient-controlled epidural analgesia using SVMand LR [J]. Scientific Reports, 2016, 6: 27041.
[41]	WHITLOCK E L, FEINER J R, CHEN L L. Perioperativemortality, 2010 to 2014: A retrospective cohortstudy using the national anesthesia clinical outcomesregistry [J]. Anesthesiology, 2015, 123(6): 1312-1321.
[42]	HOVE L D, STEINMETZ J, CHRISTOFFERSEN JK, et al. Analysis of deaths related to anesthesia in theperiod 1996-2004 from closed claims registered by theDanish Patient Insurance Association [J]. Anesthesiology,2007, 106(4): 675-680.
[43]	DETSKY M E, JIVRAJ N, ADHIKARI N K, et al.Will this patient be difficult to intubate? [J]. JAMA,2019, 321(5): 493.
[44]	CONNOR C W, SEGAL S. The importance of subjectivefacial appearance on the ability of anesthesiologiststo predict difficult intubation [J]. Anesthesia andAnalgesia, 2014, 118(2): 419-427.
[45]	CONNOR C W, SEGAL S. Accurate classification ofdifficult intubation by computerized facial analysis [J].Anesthesia and Analgesia, 2011, 112(1): 84-93.
[46]	CUENDET G L, SCHOETTKER P, Y¨UCE A, et al.Facial image analysis for fully automatic predictionof difficult endotracheal intubation [J]. IEEE Transactionson Biomedical Engineering, 2016, 63(2): 328-339.
[47]	MATAVA C, PANKIV E, AHUMADA L, et al. Artificialintelligence, machine learning and the pediatricairway [J]. Paediatric Anaesthesia, 2020, 30(3): 264-268.
[48]	DING Y M, WANG J X, GAO J D, et al. Severityevaluation of obstructive sleep apnea based on speechfeatures [J]. Sleep and Breathing, 2021, 25(2): 787-795.
[49]	ESPINOZA-CUADROS F, FERN′ANDEZ-POZO R,TOLEDANO D T, et al. Speech signal and facialimage processing for obstructive sleep apnea assessment[J]. Computational and Mathematical Methods inMedicine, 2015, 2015: 489761.
[50]	LIN C S, LI Y C, MOK M S, et al. Neural networkmodeling to predict the hypnotic effect of propofolbolus induction [C]//AMIA 2002 Annual SymposiumProceedings. San Antonio, TX: AMIA, 2002: 450-453.
[51]	IONESCU C M, DE KEYSER R, TORRICO B C,et al. Robust predictive control strategy applied forpropofol dosing using BIS as a controlled variable duringanesthesia [J]. IEEE Transactions on BiomedicalEngineering, 2008, 55(9): 2161-2170.
[52]	SEP′ULVEDA P O, CORT′INEZ L I, RECARTA, et al.Predictive ability of propofol effect-site concentrationsduring fast and slow infusion rates [J]. Acta AnaesthesiologicaScandinavica, 2010, 54(4): 447-452.
[53]	YI J M, DOH I, LEE S H, et al. Predictive performanceof a new pharmacokinetic model for propofol inunderweight patients during target-controlled infusion[J]. Acta Anaesthesiologica Scandinavica, 2019, 63(4):448-454.
[54]	NUNES C S, MENDONCA T F, AMORIM P, et al.Radial basis function neural networks versus fuzzy models to predict return of consciousness after generalanesthesia [C]//Proceedings of the 26th Annual InternationalConference of the IEEE EMBS. San Francisco,CA: IEEE, 2004: 865-868.
[55]	SANTANEN O A P, SVARTLING N, HAASIO J, etal. Neural nets and prediction of the recovery rate fromneuromuscular block [J]. European Journal of Anaesthesiology,2003, 20(2): 87-92.
[56]	NAIR A A, VELAGAPUDI M A, LANG J A, et al.Machine learning approach to predict postoperativeopioid requirements in ambulatory surgery patients [J].PLoS One, 2020, 15(7): e0236833.
[57]	LEE S, WEI S J, WHITE V, et al. Classification ofopioid usage through semi-supervised learning for totaljoint replacement patients [J]. IEEE Journal ofBiomedical and Health Informatics, 2021, 25(1): 189-200.
[58]	LU Y N, FORLENZA E, WILBUR R R, etal. Machine-learning model successfully predicts patientsat risk for prolonged postoperative opioiduse following elective knee arthroscopy [J]. KneeSurgery, Sports Traumatology, Arthroscopy, 2021.https://doi.org/10.1007/s00167-020-06421-7.
[59]	ELLIS R J, WANG Z C, GENES N, et al. Predictingopioid dependence from electronic health records withmachine learning [J]. BioData Mining, 2019, 12: 3.
[60]	JUNGQUIST C R, CHANDOLA V, SPULECKI C,et al. Identifying patients experiencing opioid-inducedrespiratory depression during recovery from anesthesia:The application of electronic monitoring devices[J]. Worldviews on Evidence-Based Nursing, 2019,16(3): 186-194.
[61]	RAHMAN Q A, JANMOHAMED T, CLARKE H, etal. Interpretability and class imbalance in predictionmodels for pain volatility in manage my pain app users:Analysis using feature selection and majority votingmethods [J]. JMIR Medical Informatics, 2019, 7(4):e15601.
[62]	HU Y J, KU T H, JAN R H, et al. Decision treebasedlearning to predict patient controlled analgesiaconsumption and readjustment [J]. BMC Medical Informaticsand Decision Making, 2012, 12: 131.
[63]	MILLER D D, BROWN E W. Artificial intelligence inmedical practice: The question to the answer? [J]. TheAmerican Journal of Medicine, 2018, 131(2): 129-133.
[64]	ALEXANDER J C, JOSHI G P. Anesthesiology, automation,and artificial intelligence [J]. Baylor UniversityMedical Center Proceedings, 2018, 31(1): 117-119.
[65]	LI W, LIU H, YANG P, et al. Supporting regularizedlogistic regression privately and efficiently [J]. PLoSOne, 2016, 11(6): e0156479.
[66]	CHAPALAIN X, HUET O. Is artificial intelligence(AI) at the doorstep of Intensive Care Units (ICU)and operating room (OR)? [J]. Anaesthesia, CriticalCare & Pain Medicine, 2019, 38(4): 337-338.
[67]	CHAR D S, SHAH N H, MAGNUS D. Implementingmachine learning in health care-addressing ethicalchallenges [J]. The New England Journal of Medicine,2018, 378(11): 981-983.