Machine learning methods for perioperative anesthetic management in cardiac surgery patients: a scoping review

Abstract

Background: Machine learning (ML) is developing fast with promising prospects within medicine and already has several applications in perioperative care. We conducted a scoping review to examine the extent and potential limitations of ML implementation in perioperative anesthetic care, specifically in cardiac surgery patients. Methods: We mapped the current literature by searching three databases: MEDLINE (Ovid), EMBASE (Ovid), and Cochrane Library. Articles were eligible if they reported on perioperative ML use in the field of cardiac surgery with relevance to anesthetic practices. Data on the applicability of ML and comparability to conventional statistical methods were extracted. Results: Forty-six articles on ML relevant to the work of the anesthesiologist in cardiac surgery were identified. Three main categories emerged: (I) event and risk prediction, (II) hemodynamic monitoring, and (III) automation of echocardiography. Prediction models based on ML tend to behave similarly to conventional statistical methods. Using dynamic hemodynamic or ultrasound data in ML models, however, shifts the potential to promising results. Conclusions: ML in cardiac surgery is increasingly used in perioperative anesthetic management. The majority is used for prediction purposes similar to conventional clinical scores. Remarkable ML model performances are achieved when using real-time dynamic parameters. However, beneficial clinical outcomes of ML integration have yet to be determined. Nonetheless, the first steps introducing ML in perioperative anesthetic care for cardiac surgery have been taken.

Risk Stratification Index 3.0, a Broad Set of Models for Predicting Adverse Events during and after Hospital Admission

“…Although interpretable models are frequently preferred even among experts, machine learning models are increasingly popular. 43 Some reports suggest better performance for ML models compared to traditional clinical scoring systems, 17,18 but few have been validated on multiple external datasets. We evaluated five of the most commonly used methods.…”

“…The conservative 0.7 minimum acceptance threshold for AUC was based on consultation with clinical advisors and a literature review indicating the acceptability of numerous perioperative machine learning models with c-statistics in the 0.7 to 0.8 range. 18,30 Because no a priori hypotheses were tested, we did not estimate required sample size, and instead used all eligible cases available in the Medicare fee-for-service files for the selected years. To evaluate the importance of endpoint-specific models, we compared incidence of various complications in patients selected for having the highest 5% risk of 90-day mortality to those with the highest 5% risk of specific complications.…”

“…However, machine learning methods have become increasingly popular and have shown better predictive perioperative performance than clinical scoring systems in some but not all settings. [17][18][19] We therefore developed analogous models using five commonly used machine learning methods and compared model performance characteristics with each method.…”

Systems Anesthesiology: Integrating Insights From Diverse Disciplines to Improve Perioperative Care

“…18 Many predictive models significantly increase in accuracy with increased data, 19 thus complex full-waveform data streams, such as intraoperative electroencephalogram (EEG) signals, heart rate, blood pressure, and end-tidal Co 2 , have been increasingly used as model inputs. [20][21][22] Real-time data streams also make possible closed-loop control systems to assist with patient care, such as titration of sedation or vasopressors. 23 Because of the challenge of complex artificial intelligence (AI) models to perform well across different institutions and time periods, traditional statistical models still have a role and can sometimes outperform deep learning models.…”

generic heart diseases prediction and application of genetic algorithms in healthcare systems

“…Proper prognostication of postoperative morbidity and mortality would be informative, precluding overestimation of risk and denial of surgery for patients deserving it, which could occur with some prediction models due to the high prevalence of comorbid conditions in an elderly individual. A short length of stay after cardiac surgery was correlated with younger age, no preoperative use of beta-blockers, shorter cross-clamp time, and absence of congestive heart failure showed that GA by applying GA can produce better predictivity in medicine [29].…”