In two-thirds of intensive care unit (ICU) patients and 90% of surgical patients, arterial blood pressure (ABP) is monitored non-invasively but intermittently using a blood pressure cuff. Since even ...a few minutes of hypotension increases the risk of mortality and morbidity, for the remaining (high-risk) patients ABP is measured continuously using invasive devices, and derived values are extracted from the recorded waveforms. However, since invasive monitoring is associated with major complications (infection, bleeding, thrombosis), the ideal ABP monitor should be both non-invasive and continuous. With large volumes of high-fidelity physiological waveforms, it may be possible today to impute a physiological waveform from other available signals. Currently, the state-of-the-art approaches for ABP imputation only aim at intermittent systolic and diastolic blood pressure imputation, and there is no method that imputes the continuous ABP waveform. Here, we developed a novel approach to impute the continuous ABP waveform non-invasively using two continuously-monitored waveforms that are currently part of the standard-of-care, the electrocardiogram (ECG) and photo-plethysmogram (PPG), by adapting a deep learning architecture designed for image segmentation. Using over 150,000 min of data collected at two separate health systems from 463 patients, we demonstrate that our model provides a highly accurate prediction of the continuous ABP waveform (root mean square error 5.823 (95% CI 5.806-5.840) mmHg), as well as the derived systolic (mean difference 2.398 ± 5.623 mmHg) and diastolic blood pressure (mean difference - 2.497 ± 3.785 mmHg) compared to arterial line measurements. Our approach can potentially be used to measure blood pressure continuously and non-invasively for all patients in the acute care setting, without the need for any additional instrumentation beyond the current standard-of-care.
The practice of anesthesiology is inextricably dependent upon technology. Anesthetics were first made possible, then increasingly safe, and now more scalable and efficient in part due to advances in ...monitoring and delivery technology. Herein, we discuss salient advances of the last three years in the technology of anesthesiology.
Consumer technology and telemedicine have exploded onto the scene of outpatient medicine, and perioperative management is no exception. Preoperative evaluations have been done via teleconference, and copious consumer-generated health data is available. Regulators have acknowledged the vast potential found in the transfer of consumer technology to medical practice, but issues of privacy, data ownership/security, and validity remain.
Inside the operating suite, monitoring has become less invasive, and clinical decision support systems are common. These technologies are susceptible to the "garbage in, garbage out" conundrum plaguing artificial intelligence, but they will improve as network latency decreases. Automation looms large in the future of anesthesiology as closed-loop anesthesia delivery systems are being tested in combination (moving toward a comprehensive system).
Moving forward, consumer health companies will search for applications of their technology, and loosely regulated health markets will see earlier adoption of next-generation technology. Innovations coming to anesthesia will need to account for human factors as the anesthesia provider is increasingly considered a component of the patient care apparatus.
The outbreak of COVID-19 has remained uncontained with urgent need for robust therapeutics. We have previously reported sex difference of COVID-19 for the first time indicating male predisposition. ...Males are more susceptible than females, and more often to develop into severe cases with higher mortality. This predisposition is potentially linked to higher prevalence of cigarette smoking. Nonetheless, we found for the first time that cigarette smoking extract (CSE) had no effect on angiotensin converting enzyme 2 (ACE2) and transmembrane protease serine 2 (TMPRSS2) expression in endothelial cells. The otherwise observed worse outcomes in smokers is likely linked to baseline respiratory diseases associated with chronic smoking. Instead, we hypothesized that estrogen mediated protection might underlie lower morbidity, severity and mortality of COVID-19 in females. Of note, endothelial inflammation and barrier dysfunction are major mediators of disease progression, and development of acute respiratory distress syndrome (ARDS) and multi-organ failure in patients with COVID-19. Therefore, we investigated potential protective effects of estrogen on endothelial cells against oxidative stress induced by interleukin-6 (IL-6) and SARS-CoV-2 spike protein (S protein). Indeed, 17β-estradiol completely reversed S protein-induced selective activation of NADPH oxidase isoform 2 (NOX2) and reactive oxygen species (ROS) production that are ACE2-dependent, as well as ACE2 upregulation and induction of pro-inflammatory gene monocyte chemoattractant protein-1 (MCP-1) in endothelial cells to effectively attenuate endothelial dysfunction. Effects of IL-6 on activating NOX2-dependent ROS production and upregulation of MCP-1 were also completely attenuated by 17β-estradiol. Of note, co-treatment with CSE had no additional effects on S protein stimulated endothelial oxidative stress, confirming that current smoking status is likely unrelated to more severe disease in chronic smokers. These data indicate that estrogen can serve as a novel therapy for patients with COVID-19 via inhibition of initial viral responses and attenuation of cytokine storm induced endothelial dysfunction, to substantially alleviate morbidity, severity and mortality of the disease, especially in men and post-menopause women. Short-term administration of estrogen can therefore be readily applied to the clinical management of COVID-19 as a robust therapeutic option.
Mechanisms underlying SARS-CoV-2 S protein and cytokine storm induced endothelial dysfunction and inflammation, and the mechanistic therapeutic application of 17β-estradiol to the treatment of COVID-19. Upon binding of S protein to its membrane receptor ACE2, S protein stimulates NOX2 dependent reactive oxygen species (ROS) production. Excessive ROS production by S protein induces ROS dependent cellular signaling including induction of cytokines and chemokines (e.g. IL-6 and MCP-1), all of which contribute to vascular inflammation. IL-6 also induces ROS production in a NOX2 dependent manner, aggravating endothelial oxidative stress, which in turn sustains endothelial dysfunction and vascular inflammation. This vicious cycle can be inhibited by 17β-estradiol (E2) treatment since it completely or substantially alleviated S protein induction of ACE2, and ACE2 dependent activation of NOX2, MCP-1 and ROS production by S protein. Estrogen treatment also attenuated IL-6 induced NOX2 activation and ROS production. These data indicate robust therapeutic effects of estrogen on SARS-CoV-2 infection and cytokine storm induced endothelial dysfunction and inflammation, leading to attenuated acute respiratory distress syndrome (ARDS)/multi-organ failure and reduced mortality, especially in men and post-menopause women. Display omitted
While deep neural networks (DNNs) and other machine learning models often have higher accuracy than simpler models like logistic regression (LR), they are often considered to be "black box" models ...and this lack of interpretability and transparency is considered a challenge for clinical adoption. In healthcare, intelligible models not only help clinicians to understand the problem and create more targeted action plans, but also help to gain the clinicians' trust. One method of overcoming the limited interpretability of more complex models is to use Generalized Additive Models (GAMs). Standard GAMs simply model the target response as a sum of univariate models. Inspired by GAMs, the same idea can be applied to neural networks through an architecture referred to as Generalized Additive Models with Neural Networks (GAM-NNs). In this manuscript, we present the development and validation of a model applying the concept of GAM-NNs to allow for interpretability by visualizing the learned feature patterns related to risk of in-hospital mortality for patients undergoing surgery under general anesthesia. The data consists of 59,985 patients with a feature set of 46 features extracted at the end of surgery to which we added previously not included features: total anesthesia case time (1 feature); the time in minutes spent with mean arterial pressure (MAP) below 40, 45, 50, 55, 60, and 65 mmHg during surgery (6 features); and Healthcare Cost and Utilization Project (HCUP) Code Descriptions of the Primary current procedure terminology (CPT) codes (33 features) for a total of 86 features. All data were randomly split into 80% for training (n = 47,988) and 20% for testing (n = 11,997) prior to model development. Model performance was compared to a standard LR model using the same features as the GAM-NN. The data consisted of 59,985 surgical records, and the occurrence of in-hospital mortality was 0.81% in the training set and 0.72% in the testing set. The GAM-NN model with HCUP features had the highest area under the curve (AUC) 0.921 (0.895-0.95). Overall, both GAM-NN models had higher AUCs than LR models, however, had lower average precisions. The LR model without HCUP features had the highest average precision 0.217 (0.136-0.31). To assess the interpretability of the GAM-NNs, we then visualized the learned contributions of the GAM-NNs and compared against the learned contributions of the LRs for the models with HCUP features. Overall, we were able to demonstrate that our proposed generalized additive neural network (GAM-NN) architecture is able to (1) leverage a neural network's ability to learn nonlinear patterns in the data, which is more clinically intuitive, (2) be interpreted easily, making it more clinically useful, and (3) maintain model performance as compared to previously published DNNs.
The perioperative surgical home model highlights the need for trainees to include modalities that are focused on the entire perioperative experience. The focus of this study was to design, introduce, ...and evaluate the integration of a whole-body point-of-care (POC) ultrasound curriculum (Focused periOperative Risk Evaluation Sonography Involving Gastroabdominal Hemodynamic and Transthoracic ultrasound) into residency training.
For 2 yr, anesthesiology residents (n = 42) received lectures using a model/simulation design and half were also randomly assigned to receive pathology assessment training. Posttraining performance was assessed through Kirkpatrick levels 1 to 4 outcomes based on the resident satisfaction surveys, multiple-choice tests, pathologic image evaluation, human model testing, and assessment of clinical impact via review of clinical examination data.
Evaluation of the curriculum demonstrated high satisfaction scores (n = 30), improved content test scores (n = 37) for all tested categories (48 ± 16 to 69 ± 17%, P < 0.002), and improvement on human model examinations. Residents randomized to receive pathology training (n = 18) also showed higher scores compared with those who did not (n = 19) (9.1 ± 2.5 vs. 17.4 ± 3.1, P < 0.05). Clinical examinations performed in the organization after the study (n = 224) showed that POC ultrasound affected clinical management at a rate of 76% and detected new pathology at a rate of 31%.
Results suggest that a whole-body POC ultrasound curriculum can be effectively taught to anesthesiology residents and that this training may provide clinical benefit. These results should be evaluated within the context of the perioperative surgical home.
Mechanical dyssynchrony is a potential means to predict response to cardiac resynchronization therapy (CRT). We hypothesized that novel echocardiographic image speckle tracking can quantify ...dyssynchrony and predict response to CRT.
Seventy-four subjects were studied: 64 heart failure patients undergoing CRT (aged 64+/-12 years, ejection fraction 26+/-6%, QRS duration 157+/-28 ms) and 10 normal controls. Speckle tracking applied to routine midventricular short-axis images calculated radial strain from multiple circumferential points averaged to 6 standard segments. Dyssynchrony from timing of speckle-tracking peak radial strain was correlated with tissue Doppler measures in 47 subjects (r=0.94, P<0.001; 95% CI 0.90 to 0.96). The ability of baseline speckle-tracking radial dyssynchrony (time difference in peak septal wall-to-posterior wall strain > or =130 ms) to predict response to CRT was then tested. It predicted an immediate increase in stroke volume in 48 patients studied the day after CRT with 91% sensitivity and 75% specificity. In 50 patients with long-term follow-up 8+/-5 months after CRT, baseline speckle-tracking radial dyssynchrony predicted a significant increase in ejection fraction with 89% sensitivity and 83% specificity. Patients in whom left ventricular lead position was concordant with the site of latest mechanical activation by speckle-tracking radial strain had an increase in ejection fraction from baseline to a greater degree (10+/-5%) than patients with discordant lead position (6+/-5%; P<0.05).
Speckle-tracking radial strain can quantify dyssynchrony and predict immediate and long-term response to CRT and has potential for clinical application.