A5046347872- Cardiac Arrest and Resuscitation
- Heart Rate Variability and Autonomic Control
- Non-Invasive Vital Sign Monitoring
- ECG Monitoring and Analysis
- Model Reduction and Neural Networks
- Time Series Analysis and Forecasting
- Fluid Dynamics and Turbulent Flows
- Advanced Image Processing Techniques
- Earthquake Detection and Analysis
- Probabilistic and Robust Engineering Design
- Structural Health Monitoring Techniques
- Adaptive optics and wavefront sensing
- Healthcare Technology and Patient Monitoring
- Traumatic Brain Injury and Neurovascular Disturbances
- Cardiac electrophysiology and arrhythmias
University of Washington Applied Physics Laboratory
2020-2023
University of Washington
2020-2021
Center for American Progress
2021
Progress (Ireland)
2020
Resuscitation Council
2020
Dynamic mode decomposition (DMD) provides a regression framework for adaptively learning best-fit linear dynamics model over snapshots of temporal, or spatio-temporal, data. A variety techniques have been developed producing the approximation whose solutions are exponentials in time. For spatio-temporal data, DMD low-rank and interpretable models form dominant modal structures along with their exponential/oscillatory behaviour The majority algorithms, however, prone to bias errors from noisy...
Current resuscitation protocols require pausing chest compressions during cardiopulmonary (CPR) to check for a pulse. However, CPR when patient is pulseless can worsen outcomes. Our objective was design and evaluate an ECG-based algorithm that predicts pulse presence with or without CPR. We evaluated 383 patients being treated out-of-hospital cardiac arrest real-time ECG, impedance audio recordings. Paired ECG segments having organized rhythm immediately preceding (during CPR) the (without...
Aero-optical beam control relies on the development of low-latency forecasting techniques to quickly predict wavefronts aberrated by turbulent boundary layer around an airborne optical system, and its study applies a multidomain need from astronomy microscopy for high-fidelity laser propagation. We leverage capabilities dynamic mode decomposition (DMD) — equation-free, data-driven method identifying coherent flow structures their associated spatiotemporal dynamics estimate future state...
We demonstrate the use of physics-informed machine learning algorithms for adaptive, real-time characterization aero-optical systems. From deep to nonlinear control methods, optical sciences are an ideal platform integrating data-driven and robust system identification. For specific case aero-optics, ability extract dominant coherent structures, transients turbulent behaviors is critical a diverse number applications, including complex dynamic aero-optic effects on airborne-based laser...
Objective: Current resuscitation protocols require pausing chest compressions during cardiopulmonary (CPR) to check for a pulse. However, CPR pulseless rhythm can worsen patient outcome. Our objective is design an ECG-based algorithm that predicts pulse status uninterrupted and evaluate its performance. Methods: We evaluated 383 patients being treated out-of-hospital cardiac arrest using defibrillator data. collected paired immediately adjacent ECG segments having organized rhythm. Segments...
Introduction: Cardiac arrest resuscitation requires CPR interruption for ECG rhythm analysis, but pausing is adversely associated with survival. Ideally, automated analysis would occur agnostic of state throughout and discriminate non-shockable from shockable rhythms. Transfer learning pre-trained deep convolutional neural networks (CNNs) may enable accurate when applied to time-frequency representations the ECG. We designed evaluated a transfer algorithm identify ventricular fibrillation...
Introduction: Chest compressions (CCs) during CPR cause electrical artifacts in the ECG. Prior work has found that severity of CC artifact, quantified by signal-to-noise ratio (SNR), affects diagnostic sensitivity defibrillator algorithms designed to detect shockable rhythms CCs. Whether SNR is altered defibrillation unknown. We therefore compared before and after shocks. Methods: evaluated patients with out-of-hospital cardiac arrest who received at least 1 shock, had subsequent ventricular...
Background: Current resuscitation protocols require pausing cardiopulmonary (CPR) to check for a pulse. However, CPR during pulseless rhythm worsens patient outcome. We designed an ECG-based algorithm that predicts pulse status uninterrupted CPR, and evaluated its performance both with without CPR. Methods: 383 patients who were treated out-of-hospital cardiac arrest using defibrillators real-time ECG, audio recordings. collected paired adjacent ECG segments organized rhythms. Segments the...
Background: Resuscitation from out-of-hospital cardiac arrest (OHCA) due to ventricular fibrillation (VF) typically involves continuous CPR cycles interrupted every 2 minutes for rhythm analysis and potential defibrillation. Quantitative measures of the VF ECG waveform have been proposed guide therapy because they are associated with myocardial energetics, dynamic over course resuscitation, predict outcome. However, while until recently required interruption accurately gauge prognostic...