Implementation and prospective performance evaluation of an intraoperative duration prediction model using high throughput real-time data.

Background: Accurate real-time prediction of intraoperative duration can contribute to improved perioperative outcomes. We implemented a data pipeline for extraction of real-time data from nascent anaesthesia records and silently deployed a predictive machine learning (ML) algorithm. Methods: Clinical variables were retrieved from the electronic health record via a third-party clinical decision support platform and contemporaneously ingested into a previously developed ML model. The model was trained using 3 months data, and performance was subsequently evaluated over 10 months using continuous ranked probability score. Results: The ML model made 6 173 435 predictions on 62 142 procedures. Mean continuous ranked probability score for the ML model was 27.19 (standard error 0.016) min compared with 51.66 (standard error 0.029) min for the bias-corrected scheduled duration. Linear regression did not demonstrate performance drift over the testing period. Conclusions: We implemented and silently deployed a real-time ML algorithm for predicting surgery duration. Prospective evaluation showed that model performance was preserved over a 10-month testing period.

Outcomes in pediatric liver transplant recipients receiving bilateral continuous erector spinae plane blocks.

BACKGROUND: Regional anesthesia allows for opioid-sparing and enhanced recovery after many major surgeries. Erector spinae blockade, with reduced bleeding risk and the option for continuous infusion, offers an opportunity to promote this principle in pediatric liver transplant patients. Our goal was to evaluate pain scores, opioid use, and return of bowel function following continuous ESP blockade in pediatric liver transplant recipients. METHODS: This retrospective cohort study included extubated patients who received a liver transplant at St. Louis Children's Hospital from July 2016 to July 2021. The control group, which did not meet the criteria for ESP blockade and received standard analgesia regimens, was compared to the group receiving continuous ESP blockade. Measured outcomes included pain scores, opioid consumption through postoperative day two, date of first bowel movement, and length of stay in the ICU and the hospital. RESULTS: Patient demographics between control and ESP groups showed no significant differences. Pain scores between control and ESP groups also showed no significant differences. Intraoperative and postoperative opioid requirements, studied in oral morphine equivalents per kilogram (OME/kg), were significantly lower for patients with ESP blockade. Time to first bowel movement was also significantly earlier for the ESP group. No significant differences were found in length of ICU or hospital stay. There were no safety concerns or complications related to ESP blockade. CONCLUSIONS: Use of continuous ESP blockade resulted in reduced opioid consumption through postoperative day two and earlier return of bowel function.

Continuous real-time prediction of surgical case duration using a modular artificial neural network.

BACKGROUND: Real-time prediction of surgical duration can inform perioperative decisions and reduce surgical costs. We developed a machine learning approach that continuously incorporates preoperative and intraoperative information for forecasting surgical duration. METHODS: Preoperative (e.g. procedure name) and intraoperative (e.g. medications and vital signs) variables were retrieved from anaesthetic records of surgeries performed between March 1, 2019 and October 31, 2019. A modular artificial neural network was developed and compared with a Bayesian approach and the scheduled surgical duration. Continuous ranked probability score (CRPS) was used as a measure of time error to assess model accuracy. For evaluating clinical performance, accuracy for each approach was assessed in identifying cases that ran beyond 15:00 (commonly scheduled end of shift), thus identifying opportunities to avoid overtime labour costs. RESULTS: The analysis included 70 826 cases performed at eight hospitals. The modular artificial neural network had the lowest time error (CRPS: mean=13.8; standard deviation=35.4 min), which was significantly better (mean difference=6.4 min [95% confidence interval: 6.3-6.5]; P<0.001) than the Bayesian approach. The modular artificial neural network also had the highest accuracy in identifying operating theatres that would overrun 15:00 (accuracy at 1 h prior=89%) compared with the Bayesian approach (80%) and a naïve approach using the scheduled duration (78%). CONCLUSIONS: A real-time neural network model using preoperative and intraoperative data had significantly better performance than a Bayesian approach or scheduled duration, offering opportunities to avoid overtime labour costs and reduce the cost of surgery by providing superior real-time information for perioperative decision support.

Probabilistic forecasting of surgical case duration using machine learning: model development and validation.

OBJECTIVE: Accurate estimations of surgical case durations can lead to the cost-effective utilization of operating rooms. We developed a novel machine learning approach, using both structured and unstructured features as input, to predict a continuous probability distribution of surgical case durations. MATERIALS AND METHODS: The data set consisted of 53 783 surgical cases performed over 4 years at a tertiary-care pediatric hospital. Features extracted included categorical (American Society of Anesthesiologists [ASA] Physical Status, inpatient status, day of week), continuous (scheduled surgery duration, patient age), and unstructured text (procedure name, surgical diagnosis) variables. A mixture density network (MDN) was trained and compared to multiple tree-based methods and a Bayesian statistical method. A continuous ranked probability score (CRPS), a generalized extension of mean absolute error, was the primary performance measure. Pinball loss (PL) was calculated to assess accuracy at specific quantiles. Performance measures were additionally evaluated on common and rare surgical procedures. Permutation feature importance was measured for the best performing model. RESULTS: MDN had the best performance, with a CRPS of 18.1 minutes, compared to tree-based methods (19.5-22.1 minutes) and the Bayesian method (21.2 minutes). MDN had the best PL at all quantiles, and the best CRPS and PL for both common and rare procedures. Scheduled duration and procedure name were the most important features in the MDN. CONCLUSIONS: Using natural language processing of surgical descriptors, we demonstrated the use of ML approaches to predict the continuous probability distribution of surgical case durations. The more discerning forecast of the ML-based MDN approach affords opportunities for guiding intelligent schedule design and day-of-surgery operational decisions.

Early experiences with the use of continuous erector spinae plane blockade for the provision of perioperative analgesia for pediatric liver transplant recipients.

OBJECTIVE: Pediatric liver transplantation presents a number of anesthetic challenges, especially in providing adequate perioperative analgesia. In an effort to reduce opioid consumption and improve functional outcomes following pediatric liver transplantation, we have instituted a novel analgesia protocol centered on the provision of continuous regional analgesia with erector spinae plane (ESP) blockade. CASES: We describe preincisional bilateral ESP catheter placement in two pediatric patients undergoing orthotopic liver transplantation. The first case was a 12-year-old boy with maple syrup urine disease undergoing initial transplantation and the second case was an 8-year-old boy who underwent an 11 hours complex redo liver transplant in the setting of glycogen storage disease type 1A requiring initial liver transplant in 2014. The 8-year-old boy presented to the operating suite with acute Budd-Chiari syndrome with comorbid ascites and a large right pleural effusion. In both cases, ESP blockade resulted in good analgesia, markedly reduced intraoperative and postoperative opioid consumption as compared with institutional data and published rates of consumption and was associated with rapid return of bowel function. CONCLUSIONS: These early experiences suggest a role for continuous ESP blockade to improve analgesia and potentially change the paradigm of treatment in this fragile patient population. The technique should be considered in similar interventions. Further study will be undertaken to validate our observation.

Continuous Erector Spinae Plane Catheter for Analgesia After Infant Thoracotomy: A Case Report.

The erector spinae plane block is an emerging technique for the provision of thoracolumbar analgesia with reported pediatric applications. We describe the placement of a continuous erector spinae plane catheter at the T5-T6 level in a 7-month-old infant who was undergoing thoracotomy for left upper lobectomy due to congenital pulmonary airway malformation. This technique resulted in outstanding analgesia without the need for opioid rescue analgesia. This block has a low degree of technical difficulty and is placed in area devoid of nearby critical structures and could be used in a number of potentially painful interventions.

Quantification of Postmembrane Gaseous Microembolization During Venoarterial Extracorporeal Membrane Oxygenation.

Prolonged use of venoarterial extracorporeal membrane oxygenation (VA ECMO) may be complicated by end-organ dysfunction. Although gaseous microemboli (GME) are thought to damage end organs during cardiopulmonary bypass, patient exposures to GME have not been well characterized during VA ECMO. We therefore performed an observational study of GME in adult VA ECMO patients, with correlation to clinical events during routine patient care. After institutional review board (IRB) approval, we used two Doppler probes to detect GME noninvasively in extracorporeal membrane oxygenation (ECMO) circuits on four patients for 15 hours total while also recording patient care events. We then conducted in vitro trials to compare Doppler signals with gold-standard measurements using an Emboli Detection and Classification EDAC quantifier (Luna Innnovations, Inc. Roanoke, VA) (Terumo Cardiovascular, Ann Arbor, MI) during simulated clinical interventions. Correlations between Doppler and EDAC data were used to estimate GME counts and volumes represented by clinical Doppler data. A total of 503 groups of Doppler peaks representing GME showers were observed, including 194 statistically larger showers during patient care activities containing 92% of total Doppler peaks. Intravenous injections accounted for an estimated 68% of GME and 88% of GME volume, whereas care involving movement accounted for an estimated 6% of GME and 3% of volume. Overall estimated embolic rates of 24,000 GME totaling 4 µl/hr rivals reported GME rates during cardiopulmonary bypass. Numerous GME are present in the postmembrane circuit during VA ECMO, raising concern for effects on microcirculation and organ dysfunction. Strategies to detect and minimize GME may be warranted to limit embolic exposures experienced by VA ECMO patients.

Systematic identification of gene activities promoting hypoxic death.

The sensitivity of an organism to hypoxic injury varies widely across species and among cell types. However, a systematic description of the determinants of metazoan hypoxic sensitivity is lacking. Toward this end, we screened a whole-genome RNAi library for genes that promote hypoxic sensitivity in Caenorhabditis elegans. RNAi knockdown of 198 genes conferred an invariant hypoxia-resistant phenotype (Hyp-r). Eighty-six per cent of these hyp genes had strong homologs in other organisms, 73 with human reciprocal orthologs. The hyp genes were distributed among multiple functional categories. Transcription factors, chromatin modifying enzymes, and intracellular signaling proteins were highly represented. RNAi knockdown of about half of the genes produced no apparent deleterious phenotypes. The hyp genes had significant overlap with previously identified life span extending genes. Testing of the RNAi's in a mutant background defective in somatic RNAi machinery showed that most genes function in somatic cells to control hypoxic sensitivity. DNA microarray analysis identified a subset of the hyp genes that may be hypoxia regulated. siRNA knockdown of human orthologs of the hyp genes conferred hypoxia resistance to transformed human cells for 40% of the genes tested, indicating extensive evolutionary conservation of the hypoxic regulatory activities. The results of the screen provide the first systematic picture of the genetic determinants of hypoxic sensitivity. The number and diversity of genes indicates a surprisingly nonredundant genetic network promoting hypoxic sensitivity.