Forecasting algorithms in the ICU.

Despite significant advances in modeling methods and access to large datasets, there are very few real-time forecasting systems deployed in highly monitored environment such as the intensive care unit. Forecasting models may be developed as classification, regression or time-to-event tasks; each could be using a variety of machine learning algorithms. An accurate and useful forecasting systems include several components beyond a forecasting model, and its performance is assessed using end-user-centered metrics. Several barriers to implementation and acceptance persist and clinicians will play an active role in the successful deployment of this promising technology.

Forecasting imminent atrial fibrillation in long-term electrocardiogram recordings.

BACKGROUND: Despite the morbidity associated with acute atrial fibrillation (AF), no models currently exist to forecast its imminent onset. We sought to evaluate the ability of deep learning to forecast the imminent onset of AF with sufficient lead time, which has important implications for inpatient care. METHODS: We utilized the Physiobank Long-Term AF Database, which contains 24-h, labeled ECG recordings from patients with a history of AF. AF episodes were defined as ≥5 min of sustained AF. Three deep learning models incorporating convolutional and transformer layers were created for forecasting, with two models focusing on the predictive nature of sinus rhythm segments and AF epochs separately preceding an AF episode, and one model utilizing all preceding waveform as input. Cross-validated performance was evaluated using area under time-dependent receiver operating characteristic curves (AUC(t)) at 7.5-, 15-, 30-, and 60-min lead times, precision-recall curves, and imminent AF risk trajectories. RESULTS: There were 367 AF episodes from 84 ECG recordings. All models showed average risk trajectory divergence of those with an AF episode from those without ∼15 min before the episode. Highest AUC was associated with the sinus rhythm model [AUC = 0.74; 7.5-min lead time], though the model using all preceding waveform data had similar performance and higher AUCs at longer lead times. CONCLUSIONS: In this proof-of-concept study, we demonstrated the potential utility of neural networks to forecast the onset of AF in long-term ECG recordings with a clinically relevant lead time. External validation in larger cohorts is required before deploying these models clinically.

Prediction of extubation failure in the paediatric cardiac ICU using machine learning and high-frequency physiologic data.

BACKGROUND: Cardiac intensivists frequently assess patient readiness to wean off mechanical ventilation with an extubation readiness trial despite it being no more effective than clinician judgement alone. We evaluated the utility of high-frequency physiologic data and machine learning for improving the prediction of extubation failure in children with cardiovascular disease. METHODS: This was a retrospective analysis of clinical registry data and streamed physiologic extubation readiness trial data from one paediatric cardiac ICU (12/2016-3/2018). We analysed patients' final extubation readiness trial. Machine learning methods (classification and regression tree, Boosting, Random Forest) were performed using clinical/demographic data, physiologic data, and both datasets. Extubation failure was defined as reintubation within 48 hrs. Classifier performance was assessed on prediction accuracy and area under the receiver operating characteristic curve. RESULTS: Of 178 episodes, 11.2% (N = 20) failed extubation. Using clinical/demographic data, our machine learning methods identified variables such as age, weight, height, and ventilation duration as being important in predicting extubation failure. Best classifier performance with this data was Boosting (prediction accuracy: 0.88; area under the receiver operating characteristic curve: 0.74). Using physiologic data, our machine learning methods found oxygen saturation extremes and descriptors of dynamic compliance, central venous pressure, and heart/respiratory rate to be of importance. The best classifier in this setting was Random Forest (prediction accuracy: 0.89; area under the receiver operating characteristic curve: 0.75). Combining both datasets produced classifiers highlighting the importance of physiologic variables in determining extubation failure, though predictive performance was not improved. CONCLUSION: Physiologic variables not routinely scrutinised during extubation readiness trials were identified as potential extubation failure predictors. Larger analyses are necessary to investigate whether these markers can improve clinical decision-making.

Influence of Early Extubation Location on Outcomes Following Pediatric Cardiac Surgery.

OBJECTIVES: Early extubation following pediatric cardiac surgery is common, but debate exists whether location affects outcome, with some centers performing routine early extubations in the operating room (odds ratio) and others in the cardiac ICU. We aimed to define early extubation practice variation across hospitals and assess impact of location on hospital length-of-stay and other outcomes. DESIGN: Secondary analysis of the Pediatric Cardiac Critical Care Consortium registry. SETTING: Twenty-eight Pediatric Cardiac Critical Care Consortium hospitals. PATIENTS: Patients undergoing Society of Thoracic Surgeons-European Association for Cardio-Thoracic Surgery mortality category 1-3 operations between August 2014 and February 2018. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We defined early extubation as extubation less than 6 hours after postoperative admission. Hospitals were categorized based on the proportion of their early extubation patients who underwent an odds ratio extubation. Categories included low- (< 50% of early extubation, n = 12), medium- (50%-90%, n = 8), or high- (> 90%, n = 8) frequency odds ratio early extubation centers. The primary outcome of interest was postoperative hospital length-of-stay. We analyzed 16,594 operations (9,143 early extubation, 55%). Rates of early extubation ranged from 16% to 100% across hospitals. Odds ratio early extubation rates varied from 16% to 99%. Patient characteristics were similar across hospital odds ratio early extubation categories. Early extubation rates paralleled the hospital odds ratio early extubation rates-77% patients underwent early extubation at high-frequency odds ratio extubation centers compared with 39% at low-frequency odds ratio extubation centers (p < 0.001). High- and low-frequency odds ratio early extubation hospitals had similar length-of-stay, cardiac arrest rates, and low mortality. However, high-frequency odds ratio early extubation hospitals used more noninvasive ventilation than low-frequency hospitals (15% vs. 9%; p < 0.01), but had fewer extubation failures (3.6% vs. 4.5%; p = 0.02). CONCLUSIONS: Considerable variability exists in early extubation practices after low- and moderate-complexity pediatric cardiac surgery. In this patient population, hospital length-of-stay did not differ significantly between centers with different early extubation strategies based on location or frequency.

Extubation Failure Rates After Pediatric Cardiac Surgery Vary Across Hospitals.

OBJECTIVES: Many hospitals aim to extubate children early after cardiac surgery, yet it remains unclear how this practice associates with extubation failure. We evaluated adjusted extubation failure rates and duration of postoperative mechanical ventilation across hospitals and assessed cardiac ICU organizational factors associated with extubation failure. DESIGN: Secondary analysis of the Pediatric Cardiac Critical Care Consortium clinical registry. SETTING: Pediatric Cardiac Critical Care Consortium cardiac ICUs. PATIENTS: Patients with qualifying index surgical procedures from August 2014 to June 2017. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We modeled hospital-level adjusted extubation failure rates using multivariable logistic regression. A previously validated Pediatric Cardiac Critical Care Consortium model was used to calculate adjusted postoperative mechanical ventilation. Observed-to-expected ratios for both metrics were derived for each hospital to assess performance. Hierarchical logistic regression was used to assess the association between cardiac ICU factors and extubation failure. Overall, 16,052 surgical hospitalizations were analyzed. Predictors of extubation failure (p < 0.05 in final case-mix adjustment model) included younger age, underweight, greater surgical complexity, airway anomaly, chromosomal anomaly/syndrome, longer cardiopulmonary bypass time, and other preoperative comorbidities. Three hospitals were better-than-expected outliers for extubation failure (95% CI around observed-to-expected < 1), and three hospitals were worse-than-expected (95% CI around observed-to-expected > 1). Two hospitals were better-than-expected outliers for both extubation failure and postoperative mechanical ventilation, and three were worse-than-expected for both. No hospital was an outlier in opposite directions. Greater nursing hours per patient day and percent nursing staff with critical care certification were associated with lower odds of extubation failure. Cardiac ICU factors such as fewer inexperienced nurses, greater percent critical care trained attendings, cardiac ICU-dedicated respiratory therapists, and fewer patients per cardiac ICU attending were not associated with lower odds of extubation failure. CONCLUSIONS: We saw no evidence that hospitals trade higher extubation failure rates for shorter duration of postoperative mechanical ventilation after pediatric cardiac surgery. Increasing specialized cardiac ICU nursing hours per patient day may achieve better extubation outcomes and mitigate the impact of inexperienced nurses.

CYP2C9*2 is associated with indomethacin treatment failure for patent ductus arteriosus.

Aims: To identify clinical andgenetic factors associated with indomethacin treatment failure in preterm neonates with patent ductus arteriosus (PDA). Patients & Methods: This is a multicenter cohort study of 144 preterm infants (22-32 weeks gestational age) at three centers who received at least one treatment course of indomethacin for PDA. Indomethacin failure was defined as requiring subsequent surgical intervention. Results: In multivariate analysis, gestational age (AOR 0.76, 95% CI 0.60-0.96), surfactant use (AOR 9.77, 95% CI 1.15-83.26), and CYP2C9*2 (AOR 3.74; 95% CI 1.34-10.44) were each associated with indomethacin failure. Conclusion: Age, surfactant use, and CYP2C9*2 influence indomethacin treatment outcome in preterm infants with PDA. This combination of clinical and genetic factors may facilitate targeted indomethacin use for PDA.