Clifton W. Callaway

The alpha-2-adrenergic agonist drug dexmedetomidine induces sleep-like sedation and reduces global metabolic rate. Drug-induced metabolic suppression sustained over days in humans could be a strategy for reducing energy expenditure (EE), oxygen consumption (VO ₂ ), and carbon dioxide production (VCO ₂ ) during prolonged rescue or austere conditions with limited life-support. We tested the feasibility of reducing EE, VO ₂ and VCO ₂ measured with intermittent indirect calorimetry using subcutaneous infusion of dexmedetomidine 1 µg/kg/hour for 18 hours per day over five days in healthy adults. Five participants received multiple days of dexmedetomidine in a laboratory setting, and five separate participants received placebo for one or five days. Relative to resting metabolic rate, dexmedetomidine infusion reduced VO ₂ (effects, 95%CI, at 6-hour time epoch: −0.084 L/min, −0.145 to −0.023), VCO ₂ (−0.067 L/min, −0.130 to −0.005) and EE (−577 kcal/day, −1016 to −137). Relative to placebo, dexmedetomidine produced light sedation (Richmond Agitation Sedation Score: −1.25; 95%CI −1.70 to −0.80 relative to placebo), with lower heart rate (−15 beats/minute; 95%CI −23 to −7) and core body temperature (−1.3ºC; 95%CI −1.7 to −0.8). Dexmedetomidine effects on all physiological variables decreased over time and were absent after day 2. Maximum and minimum plasma drug levels did not change over days. Participants reported rebound anxiety and insomnia after the study. Administering a sedating level of dexmedetomidine that reduces VO ₂ , VCO ₂ , and EE for 2 days was feasible in healthy adults. However, tolerance to drug effects develops by 3 days, which is a critical limitation for long-term use of this drug regimen.

Aims: AI in electrocardiography (ECG) has diverged into two paths: traditional signal processing with machine learning, and deep learning of raw waveforms. The former preserves physiological interpretability but may miss novel patterns, while the latter overlooks a century of ECG science. No framework yet integrates both. This study introduces a fusion model combining handcrafted feature-based random forest and waveform-based convolutional neural network to improve the diagnosis of acute coronary events. Methods and results: This prospective, multicentre cohort study included pre-hospital ECGs recorded on emergency medical services from patients with a chief complaint of chest pain. Using this dataset, we developed four fusion frameworks that integrate ECG features with waveform information: (1) decision-level fusion combining predictions from random forest and convolutional neural network (CNN); (2) decision-level fusion with retraining; (3) feature-level fusion by concatenating CNN embeddings with ECG features; and (4) feature-level fusion combining CNN embeddings with tree-level outputs from the random forest. This study comprised of 10 393 ECGs from 7397 patients. On both outcomes, acute coronary syndrome and occlusion myocardial infarction, the proposed decision fusion with retraining (Approach 2) achieved an AUROC of 0.878 (95% CI, 0.837–0.915) and 0.961 (95% CI, 0.934–0.983) on the test set and an average precision score of 0.749 (95% CI, 0.681–0.810) and 0.850 (95% CI, 0.763–0.918), respectively, outperforming all other proposed fusion and individual models. Conclusion: This study demonstrates the significance of fusion models that integrate complementary ECG representations to improve diagnostic accuracy in acute coronary events, highlighting a promising approach for real-time AI-augmented cardiovascular disease detection across diverse use cases.

Objective: Acute coronary syndrome (ACS) is a life-threatening condition requiring accurate diagnosis for better outcomes. However, variability in signs and symptoms among racial subgroups could cause disparities in diagnostic accuracy. In this study, we use machine learning models to diagnose ACS, focusing on mitigating disparities and ensuring fairness between Black and non-Black populations. Methods: We built on a state-of-the-art random forest classifier to compare three mitigation strategies. The first two approaches involved resampling or partitioning the data prior to training, while the third approach proposed an innovative framework called adversarial debiasing. To evaluate our model performance, we used the receiver operating characteristic (ROC) curve and an operating point at 80% specificity for clinical importance. Results: After mitigation with adversarial debiasing, the difference in sensitivities between the two subgroups decreased from 9.8% to 1.3%. Specifically, this approach achieved areas under the ROC of 0.810 and 0.817, and sensitivities of 70.1% and 71.4%, respectively for Black and non-Black subgroups. Conclusion: The proposed adversarial debiasing model outperformed the other two methods in both diagnostic accuracy and effectiveness in minimizing disparities. Significance: We expect this framework to achieve fair diagnostic models across diverse demographic populations globally and be generalizable to other outcomes.

Withdrawal of life sustaining treatment (WLST) is common in critically ill patients and confers a high probability of in-hospital death. In clinical trials, WLST is a specific example of a post-randomization intervention with a strong influence on outcome that can bias estimates of treatment effect. Only under several strong assumptions do the observed effects from randomized treatment in the presence of WLST correspond to those expected in its absence. However, WLST is rarely accounted for in trials. We propose a systematic approach to analyze the rates of WLST and characteristics of trial participants who died after WLST in order to identify when bias is present, and sensitivity analyses to set bounds on the direction and magnitude of this bias. In an example randomized trial of treatments for out-of-hospital cardiac arrest, the bias attributable to WLST reduces the observed treatment effects. Sensitivity analyses set bounds on this bias and estimate hypothetical treatment effects that might have been observed in the absence of WLST. This approach provides better insight into the observed results of the trial.

Objective: In this paper we develop and evaluate ECG-SMART-NET for occlusion myocardial infarction (OMI) identification. OMI is a severe form of heart attack characterized by complete blockage of one or more coronary arteries requiring immediate referral for cardiac catheterization to restore blood flow to the heart. Two thirds of OMI cases are difficult to visually identify from a 12-lead electrocardiogram (ECG) and can be potentially fatal if not identified quickly. Previous works on this topic are scarce, and current state-of-the-art evidence suggests both feature-based random forests and convolutional neural networks (CNNs) are promising approaches to improve ECG detection of OMI. Methods: While the ResNet architecture has been adapted for use with ECG recordings, it is not ideally suited to capture informative temporal features within each lead and the spatial concordance or discordance across leads. We propose a clinically informed modification of the ResNet-18 architecture. The model first learns temporal features through temporal convolutional layers with 1xk kernels followed by a spatial convolutional layer, after the residual blocks, with 12x1 kernels to learn spatial features. Results: ECG-SMART-NET was benchmarked against the original ResNet-18 and other state-of-the-art models on a multisite real-word clinical dataset that consists of 10, 393 ECGs from 7, 397 unique patients (rate of OMI = 7.2%). ECG-SMART-NET outperformed other models in the classification of OMI with a test AUC of 0.953 [0.921, 0.978]. Conclusion and Significance: ECG-SMART-NET can outperform the state-of-the-art random forest for OMI prediction and is better suited for this task than the original ResNet-18 architecture.

Background and Aims: The importance of risk stratification in patients with chest pain extends beyond diagnosis and immediate treatment. This study sought to evaluate the prognostic value of electrocardiogram feature-based machine learning models to risk-stratify all-cause mortality in those with chest pain. Methods: This was a prospective observational cohort study of consecutive, non-traumatic patients with chest pain. All-cause death was ascertained from multiple sources, including the CDC National Death Index registry. Six machine learning models were trained for survival analysis using 73 morphological electrocardiogram features (80% training with 10-fold cross-validation and 20% testing), followed by a variational Bayesian Gaussian mixture model to define distinct risk groups. The resulting classification performance was compared against the HEART score. Results: The derivation cohort included 4015 patients (age 59 ± 16 years, 47% women). The mortality rate was 20.3% after a median follow-up period of 3.05 years (interquartile range 1.75–5.32). Extra Survival Trees outperformed other forecasting models, and the derived risk groups successfully classified patients into low-, moderate-, and high-risk groups (log-rank test statistic = 121.14, P < .001). This model outperformed the HEART score, reducing the rate of missed events by >90% with a negative predictive value and sensitivity of 93.4% and 85.9%, compared to 89.0% and 75.0%, respectively. In an independent external testing cohort (N = 3095, age 59 ± 15 years, 44% women, 30-day mortality 3.5%), patients in the moderate [odds ratio 3.62 (1.35–9.74)] and high [odds ratio 6.12 (2.38–15.75)] risk groups had significantly higher odds of mortality compared to those in the low-risk group. Conclusions: The externally validated machine learning-based model, exclusively utilizing features from the 12-lead electrocardiogram, outperformed the HEART score in stratifying the mortality risk of patients with acute chest pain. This may have the potential to impact the precision of care delivery and the allocation of resources to those at highest risk of adverse events.

Background: Out‐of‐hospital cardiac arrest (OHCA) is a leading cause of morbidity and mortality. Temperature management (TM) is recommended since hyperthermia is associated with worse outcomes. Pneumonia is a frequent occurrence following OHCA, and some studies suggest that TM may have a negative impact on its development. Selective digestive decontamination (SDD) is used in some centers to reduce the incidence of pneumonia in intensive care unit (ICU), but its use remains controversial. This study aims to assess the incidence, risk factors and clinical course of VAP after OHCA. Methods: We conducted a retrospective cohort study on 169 consecutive OHCA patients after their admission in ICU. All patients were treated with TM and SDD. Pharyngeal swabs were analyzed twice weekly. The primary outcome was the incidence of VAP and non‐VAP. Secondary aim was to identify the risk factors associated with VAP and its effect on patients’ outcome. Results: Incidence of VAP was 5.3%, while incidence of non‐VAP was 9.5%. In multivariate analysis, male gender (sHR 3.01; CI 1.1–7.9), increase of white blood cells (WBC) count > 30% over 5 days (sHR 2.32; CI 1.23–3.9), heart disease (sHR 2.4; CI 1.36–4.59), and bacterial colonization of the pharynx (sHR 2.79; CI 1.13–4.39) were significantly associated with VAP. Conclusions: Pharyngeal colonization could be useful to identify patients at higher risk of VAP development. The low rate of VAP in this cohort suggests that SDD can prevent VAP after OHCA. Further studies are needed to explore the potential of SDD in OHCA patients.

Intravenous alpha-2-adrenergic receptor agonists reduce energy expenditure and lower the temperature when shivering begins in humans, allowing a decrease in core body temperature. Because there are few data about similar effects from oral drugs, we tested whether single oral doses of the sedative dexmedetomidine (1 µg/kg sublingual or 4 µg/kg swallowed) or the muscle relaxant tizanidine (8 mg or 16 mg), combined with surface cooling, reduce energy expenditure and core body temperature in humans. A total of 26 healthy participants completed 41 one-day laboratory studies measuring core body temperature using an ingested telemetry capsule and measuring energy expenditure using indirect calorimetry for up to 6 hours after drug ingestion. Dexmedetomidine induced a median 13% – 19% peak reduction and tizanidine induced a median 15% – 22% peak reduction in energy expenditure relative to baseline. Core body temperature decreased a median of 0.5°C – 0.6°C and 0.5°C – 0.7°C respectively. Decreases in temperature occurred after peak reductions in energy expenditure. Energy expenditure increased with a decrease in core temperature in control participants but did not occur after 4 µg/kg dexmedetomidine or 16 mg tizanidine. Plasma levels of dexmedetomidine but not tizanidine were related to mean temperature change. Decreases in heart rate, blood pressure, respiratory rate, cardiac stroke volume index, and cardiac index were associated with the change in metabolic rate after higher drug doses. We conclude that both oral dexmedetomidine and oral tizanidine reduce energy expenditure and allow decrease in core temperature in humans.

Objective We hypothesized that the administration of amantadine would increase awakening of comatose patients resuscitated from cardiac arrest. Methods We performed a prospective, randomized, controlled pilot trial, randomizing subjects to amantadine 100 mg twice daily or placebo for up to 7 days. The study drug was administered between 72 and 120 hours after resuscitation and patients with absent N20 cortical responses, early cerebral edema, or ongoing malignant electroencephalography patterns were excluded. Our primary outcome was awakening, defined as following two-step commands, within 28 days of cardiac arrest. Secondary outcomes included length of stay, awakening, time to awakening, and neurologic outcome measured by Cerebral Performance Category at hospital discharge. We compared the proportion of subjects awakening and hospital survival using Fisher exact tests and time to awakening and hospital length of stay using Wilcoxon rank sum tests. Results After 2 years, we stopped the study due to slow enrollment and lapse of funding. We enrolled 14 subjects (12% of goal enrollment), seven in the amantadine group and seven in the placebo group. The proportion of patients who awakened within 28 days after cardiac arrest did not differ between amantadine (n=2, 28.6%) and placebo groups (n=3, 42.9%; P>0.99). There were no differences in secondary outcomes. Study medication was stopped in three subjects (21.4%). Adverse events included a recurrence of seizures (n=2; 14.3%), both of which occurred in the placebo group.