Machine learning based analysis and detection of trend outliers for electromyographic neuromuscular monitoring.

PURPOSE: Neuromuscular monitoring is frequently plagued by artefacts, which along with the frequent unawareness of the principles of this subtype of monitoring by many clinicians, tends to lead to a cynical attitute by clinicians towards these monitors. As such, the present study aims to derive a feature set and evaluate its discriminative performance for the purpose of Train-of-Four Ratio (TOF-R) outlier analysis during continuous intraoperative EMG-based neuromuscular monitoring. METHODS: Patient data was sourced from two devices: (1) Datex-Ohmeda Electromyography (EMG) E-NMT: a dataset derived from a prospective observational trial including 136 patients (21,891 TOF-R observations), further subdivided in two based on the type of features included; and (2) TetraGraph: a clinical case repository dataset of 388 patients (97,838 TOF-R observations). The two datasets were combined to create a synthetic set, which included shared features across the two. This process led to the training of four distinct models. RESULTS: The models showed an adequate bias/variance balance, suggesting no overfitting or underfitting. Models 1 and 2 consistently outperformed the others, with the former achieving an F1 score of 0.41 (0.31, 0.50) and an average precision score (95% CI) of 0.48 (0.35, 0.60). A random forest model analysis indicated that engineered TOF-R features were proportionally more influential in model performance than basic features. CONCLUSIONS: Engineered TOF-R trend features and the resulting Cost-Sensitive Logistic Regression (CSLR) models provide useful insights and serve as a potential first step towards the automated removal of outliers for neuromuscular monitoring devices. TRIAL REGISTRATION: NCT04518761 (clinicaltrials.gov), registered on 19 August 2020.

Neuromuscular end-point predictive capability of published rocuronium pharmacokinetic/pharmacodynamic models: An observational trial.

BACKGROUND: Objective neuromuscular monitoring remains the single most reliable method to ensure optimal perioperative neuromuscular management. Nevertheless, the prediction of clinical neuromuscular endpoints by means of Pharmacokinetic (PK) and Pharmacodynamic (PD) modelling has the potential to complement monitoring and improve perioperative neuromuscular management.s STUDY OBJECTIVE: The present study aims to assess the performance of published Rocuronium PK/PD models in predicting intraoperative Train-of-four (TOF) ratios when benchmarked against electromyographic TOF measurements. DESIGN: Observational trial. SETTING: Tertiary Belgian hospital, from August 2020 up to September 2021. PATIENTS AND INTERVENTIONS: Seventy-four patients undergoing general anaesthesia for elective surgery requiring the administration of rocuronium and subject to continuous EMG neuromuscular monitoring were included. PK/PD-simulated TOF ratios were plotted and synchronised with their measured electromyographic counterparts and their differences analysed by means of Predictive Error derivatives (Varvel criteria). MAIN RESULTS: Published rocuronium PK/PD models overestimated clinically registered TOF ratios. The models of Wierda, Szenohradszky, Cooper, Alvarez-Gomez and McCoy showed significant predictive consistency between themselves, displaying Median Absolute Performance Errors between 38% and 41%, and intra-individual differences (Wobble) between 14 and 15%. The Kleijn model outperformed the former with a lower Median Absolute Performance Error (16%, 95%CI [0.01; 57]) and Wobble (11%, 95%CI [0.01; 34]). All models displayed considerably wide 95% confidence intervals for all performance metrics, suggesting a significantly variable performance. CONCLUSIONS: Simulated TOF ratios based on published PK/PD models do not accurately predict real intraoperative TOF ratio dynamics. TRIAL REGISTRATION: NCT04518761 (clinicaltrials.gov), registered on 19 August 2020.

A survey on the availability, usage and perception of neuromuscular monitors in Europe.

PURPOSE: Neuromuscular blocking agents (NMBAs) are routinely administered to patients in a multiplicity of anesthetic settings. Absence of postoperative residual neuromuscular block is widely considered an anesthetic patient safety mandate. Despite the increasing availability of a wider range of commercial quantitative neuromuscular monitors, the availability and use of neuromuscular monitoring devices is deemed to be suboptimal even in countries with above-average health system ratings. The present study aims to more precisely characterize the perceived availability, cost sensitivity and usability of neuromuscular monitoring devices within European anesthesia departments. METHODS: A pre-registered internet-based survey assessing the availability, cost sensitivity and usability of neuromuscular monitoring devices was distributed as e-mail newsletter by the European Society of Anaesthesiology and Intensive Care (ESAIC) to all of its active full members. The survey was available online for a total of 120 days. RESULTS: Having targeted a total of 7472 ESAIC members, the survey was completed by a total of 692 anesthesiologists (9.3%) distributed across 37 different European countries. Quantitative monitors were reported to be proportionally more available than qualitative ones (87.6% vs. 62.6%, respectively), as well as in greater monitor-per-operating room ratios. Most anesthesiologists (60.5%) expressed moderate confidence in quantitative monitors, with artifactual recordings and inaccurate measurements being the most frequently encountered issues (25.9%). The commercial pricing of quantitative devices was considered more representative of a device's true value, when compared to qualitative instruments (average cost of €4.500 and €1.000 per device, respectively). CONCLUSION: The availability of quantitative NMM in European operating theaters has increased in comparison with that reported in previous decades, potentially indicating increasing monitoring rates. European anesthesiologists express moderate confidence in quantitative neuromuscular monitors, along with a sentiment of adequate pricing when compared to their qualitative counterparts. Trust in quantitative monitors is marked by caution and awareness for artifactual recordings, with a consequent expectation that developments focusing on accuracy, reliability and ergonomics of neuromuscular monitors be prioritized.

Exploratory Outlier Detection for Acceleromyographic Neuromuscular Monitoring: Machine Learning Approach.

BACKGROUND: Perioperative quantitative monitoring of neuromuscular function in patients receiving neuromuscular blockers has become internationally recognized as an absolute and core necessity in modern anesthesia care. Because of their kinetic nature, artifactual recordings of acceleromyography-based neuromuscular monitoring devices are not unusual. These generate a great deal of cynicism among anesthesiologists, constituting an obstacle toward their widespread adoption. Through outlier analysis techniques, monitoring devices can learn to detect and flag signal abnormalities. Outlier analysis (or anomaly detection) refers to the problem of finding patterns in data that do not conform to expected behavior. OBJECTIVE: This study was motivated by the development of a smartphone app intended for neuromuscular monitoring based on combined accelerometric and angular hand movement data. During the paired comparison stage of this app against existing acceleromyography monitoring devices, it was noted that the results from both devices did not always concur. This study aims to engineer a set of features that enable the detection of outliers in the form of erroneous train-of-four (TOF) measurements from an acceleromyographic-based device. These features are tested for their potential in the detection of erroneous TOF measurements by developing an outlier detection algorithm. METHODS: A data set encompassing 533 high-sensitivity TOF measurements from 35 patients was created based on a multicentric open label trial of a purpose-built accelero- and gyroscopic-based neuromuscular monitoring app. A basic set of features was extracted based on raw data while a second set of features was purpose engineered based on TOF pattern characteristics. Two cost-sensitive logistic regression (CSLR) models were deployed to evaluate the performance of these features. The final output of the developed models was a binary classification, indicating if a TOF measurement was an outlier or not. RESULTS: A total of 7 basic features were extracted based on raw data, while another 8 features were engineered based on TOF pattern characteristics. The model training and testing were based on separate data sets: one with 319 measurements (18 outliers) and a second with 214 measurements (12 outliers). The F1 score (95% CI) was 0.86 (0.48-0.97) for the CSLR model with engineered features, significantly larger than the CSLR model with the basic features (0.29 [0.17-0.53]; P<.001). CONCLUSIONS: The set of engineered features and their corresponding incorporation in an outlier detection algorithm have the potential to increase overall neuromuscular monitoring data consistency. Integrating outlier flagging algorithms within neuromuscular monitors could potentially reduce overall acceleromyography-based reliability issues. TRIAL REGISTRATION: ClinicalTrials.gov NCT03605225; https://clinicaltrials.gov/ct2/show/NCT03605225.

Forty years of neuromuscular monitoring and postoperative residual curarisation: a meta-analysis and evaluation of confidence in network meta-analysis.

BACKGROUND: The reported incidence of postoperative residual curarisation (PORC) is still unacceptably high. The capacity of intraoperative neuromuscular monitoring (NMM) to reduce the incidence of PORC has yet to be established from pooled clinical studies. We conducted a meta-analysis of data from 1979 to 2019 to reanalyse this relationship. METHODS: English language, peer-reviewed, and operation room adult anaesthesia setting articles published between 1979 and 2019 were searched for on PubMed, Cochrane Central Register of Controlled Trials, ISI-WoK, and Scopus. The primary outcome was PORC incidence as defined by an at- or post-extubation train-of-four ratio (TOFR) of lower than 0.7, 0.9, or 1.0. Additional collected variables included the duration of action of neuromuscular blocking agents (NMBAs) used, sugammadex or neostigmine use, and the technique of anaesthesia maintenance. RESULTS: Fifty-three studies (109 study arms, 12 664 patients) were included. The pooled PORC incidence associated with the use of intermediate duration NMBAs and quantitative NMM was 0.115 (95% confidence interval [CI], 0.057-0.188). This was significantly lower than the PORC rate for both qualitative NMM (0.306; 95% CI, 0.09-0.411) and no NMM (0.331; 95% CI, 0.234-0.435). Anaesthesia type did not significantly affect PORC incidence. Sugammadex use was associated with lower PORC rates. The GRADE global level of evidence was very low and the refined assessment of the network meta-analysis by means of a confidence in network meta-analysis raised concerns on within- and across-study bias. CONCLUSIONS: Quantitative NMM outperforms both subjective and no NMM monitoring in reducing PORC as defined by a TOFR of <0.9.

Acceptance of mHealth among health professionals: a case study on anesthesia practitioners.

BACKGROUND: mHealth, the practice of medicine aided by mobile devices is a growing market. Although the offer on Anesthesia applications (Apps) is quite prolific, representative formal assessments on the views of anesthesia practitioners on its use and potential place in daily practice is lacking. This survey aimed thus to cross-assess the Belgian anesthesia population on the use of smartphone Apps and peripherals. METHODS: The survey was exclusively distributed as an online anonymous questionnaire. Sharing took place via hyperlink forwarding by the Belgian Society for Anesthesia and Reanimation (BSAR) and by the Belgian Association for Regional Anesthesia (BARA) to all registered members. The first answer took place on 5 September 2018, the last on 22 January 2019. RESULTS: Three hundred forty-nine answers were obtained (26.9% corresponding to trainees, 73.1% to specialists). Anesthesiologists were positively confident that Apps and peripherals could help improve anesthesia care (57.0 and 47.9%, respectively, scored 4 or 5, in a scale from 0 to 5). Trainees were significantly more confident than specialists on both mobile Apps (71.2% and 51.8%, respectively; p = 0.001) and peripherals (77.7% and 45.1%, respectively; p = 0.09). The usefulness of Apps and Peripherals was rated 1 or below (on a 0 to 5 scale), respectively, by 9.5 and 14.6% of the total surveyed population, being specialists proportionally less confident in Smartphone peripherals than trainees (p = 0.008). Mobile apps are actively used by a significantly higher proportional number of trainees (67.0% vs. 37.3%, respectively; p = 0.000001). The preferred category of mobile Apps was dose-calculating applications (39.15%), followed by digital books (21. 1%) and Apps for active perioperative monitoring (20.0%). CONCLUSIONS: Belgian Anesthesia practitioners show a global positive attitude towards smartphone Apps and Peripherals, with trainees trending to be more confident than specialists. TRIAL REGISTRATION: ClinicalTrials.gov database Identifier: NCT03750084. Retrospectively registered on 21 November 2018.

Development and validation of an android-based application for anaesthesia neuromuscular monitoring.

Quantitative neuromuscular block (NMB) assessment is an internationally recognised necessity in anesthesia care whenever neuromuscular blocking agents are administered. Despite this, the incidence of residual neuromuscular block and its associated major respiratory morbidity and mortality remain unacceptably high considering its preventable nature. Recent surveys show that quantitative NMB assessment is not consistently employed by anesthesiologists. Availability, price and practical concerns are some of the factors determining this phenomenon. Clinically assess and validate an Android cell phone application conceived specifically for NMB Monitoring in the anesthesia setting. Twenty-two adult ASA I to III patients scheduled to undergo elective surgical procedures under general anaesthesia requiring administration of a neuromuscular blocking agent were included. After anaesthesia induction, the grade of neuromuscular block was assessed at multiple independent time-points by paired comparison of the train of four (TOF) Ratios obtained by a Stimpod™ accelerometer and the currently developed application. Accelerometric measurements were made at the patient's hand after retrograde supramaximal stimulation of the ipsilateral ulnar nerve. TOF-ratios were subjected to bias analysis with 0.001 as the a priori established clinical significance cut-off. The difference between the two methods averaged 0.0004 (95% limits of agreement: ± 0.12), with 83.3% of the differences being under 0.05. This average inter-method difference was not significantly different than the a priori hypothesized difference cut-off of 0.001 (p = 0.78). Lin's concordance correlation coefficient and Pearson's correlation were both of 0.98. The custom developed Android application proved accurate for diagnosis of residual neuromuscular block.