Parameter estimation in uncertain differential equations with exponential solutions

This paper proposes three methods to estimate the parameters in uncertain differential equations (UDEs) based on discrete observation data. The first method is designed for a class of UDEs in which their solutions have the explicit expressions of uncertainty distribution. The second method is given to solve the estimation problem through the inverse uncertainty distribution. In the third method, the unknown parameters of UDEs are estimated by the solution of the corresponding α-path. These methods are interpreted to be efficient and practical by using a popular UDE with exponential solutions and obtaining the detailed estimators of the parameters.

Abstract 336: A Machine Learning Algorithm for Detecting Pulse During Out-Of-Hospital Cardiac Arrest

Introduction: Pulse detection during out-of-hospital cardiac arrest (OHCA) is a challenge still not satisfactorily solved. An automated and accurate method for detecting pulse would reduce hands-off intervals and allow for more prompt post-cardiac arrest care. The aim of this study was to develop a method based on machine learning (ML) to detect pulse during OHCA. Materials and methods: Data were gathered from 187 OHCA patients treated by Tualatin Valley Fire & Rescue (Tigard, OR, USA) using the Philips HeartStart MRx monitor/defibrillator between 2010 and 2014. The dataset used in the study contained 1140 5-s epochs presenting organized rhythms, 792 pulse-generating rhythms (PRs) and 348 pulseless electrical activity (PEA), annotated by consensus between two clinicians and a biomedical engineer using the available clinical information and the capnography signal. The dataset was split patient-wise into training (60%) and test (40%) sets. Each epoch contained the ECG and the thoracic impedance that were first preprocessed and then used to adaptively extract the impedance circulation component (ICC). The ICC shows a small fluctuation with each effective heartbeat. A total of 7 well-known waveform features were computed from the ECG and ICC and fed as observations to a discrete observation density hidden Markov model that classified each observation as PR (pulse) or PEA (no-pulse). The training set was used to develop and optimize the method, while the test set was used to measure the performance in terms of sensitivity (PR detection) and specificity (PEA detection). This procedure was repeated 50 times to estimate the distributions of the performance metrics. Results: The method showed a mean (SD) sensitivity and specificity of 95.4%(2.2) and 91.6% (3.4), respectively. Results were slightly above those previously reported by other authors using different ML techniques. Conclusions: A method based on a discrete observation density hidden Markov model can accurately detect pulse during OHCA. Further studies with larger datasets are needed to confirm these findings.