Search

Sepsis-induced acute respiratory failure (ARF) is a serious complication with a poor prognosis. This paper presents a deep representation learningbased phenotyping method to identify distinct groups of clinical trajectories of septic patients with ARF. For this retrospective study, we created a dataset from electronic medical records (EMR) consisting of data from sepsis patients admitted to medical intensive care units who required at least 24 hours of invasive mechanical ventilation at a quarternary care academic hospital in southeast USA for the years 2016-2021. A total of N=3349 patient encounters were included in this study. Clustering Representation Learning on Incomplete Time Series Data (CRLI) algorithm was applied to a parsimonious set of EMR variables in this data set. To validate the optimal number of clusters, the K-means algorithm was used in conjunction with dynamic time warping. Our model yielded four distinct patient phenotypes that were characterized as liver dysfunction/heterogeneous, hypercapnia, hypoxemia, and multiple organ dysfunction syndrome by a critical care expert. A Kaplan-Meier analysis to compare the 28-day mortality trends exhibited significant differences (p < 0.005) between the four phenotypes. The study demonstrates the utility of our deep representation learning-based approach in unraveling phenotypes that reflect the heterogeneity in sepsis-induced ARF in terms of different mortality outcomes and severity. These phenotypes might reveal important clinical insights into an effective prognosis and tailored treatment strategies., Comment: 9 pages

Objective: Blood transfusions, crucial in managing anemia and coagulopathy in ICU settings, require accurate prediction for effective resource allocation and patient risk assessment. However, existing clinical decision support systems have primarily targeted a particular patient demographic with unique medical conditions and focused on a single type of blood transfusion. This study aims to develop an advanced machine learning-based model to predict the probability of transfusion necessity over the next 24 hours for a diverse range of non-traumatic ICU patients. Methods: We conducted a retrospective cohort study on 72,072 adult non-traumatic ICU patients admitted to a high-volume US metropolitan academic hospital between 2016 and 2020. We developed a meta-learner and various machine learning models to serve as predictors, training them annually with four-year data and evaluating on the fifth, unseen year, iteratively over five years. Results: The experimental results revealed that the meta-model surpasses the other models in different development scenarios. It achieved notable performance metrics, including an Area Under the Receiver Operating Characteristic (AUROC) curve of 0.97, an accuracy rate of 0.93, and an F1-score of 0.89 in the best scenario. Conclusion: This study pioneers the use of machine learning models for predicting blood transfusion needs in a diverse cohort of critically ill patients. The findings of this evaluation confirm that our model not only predicts transfusion requirements effectively but also identifies key biomarkers for making transfusion decisions.

Pulse transit time (PTT) has been widely used for cuffless blood pressure (BP) measurement. But, it requires more than one cardiovascular signals involving more than one sensing device. In this paper, we propose a method for continuous cuffless blood pressure measurement with the help of left ventricular ejection time (LVET). The LVET is estimated using a signal obtained through a micro-electromechanical system (MEMS)-based accelerometric sensor. The sensor acquires a seismocardiogram (SCG) signal at the chest surface, and the LVET information is extracted. Both systolic blood pressure (SBP) and diastolic blood pressure (DBP) are estimated by calibrating the system with the original arterial blood pressure values of the subjects. The proposed method is evaluated using different quantitative measures on the signals collected from ten subjects under the supine position. The performance of the proposed method is also compared with two earlier approaches, where PTT intervals are estimated from electrocardiogram (ECG)-photoplethysmogram (PPG) and SCG-PPG, respectively. The performance results clearly show that the proposed method is comparable with the state-of-the-art methods. Also, the computed blood pressure is compared with the original one, measured through a CNAP system. It gives the mean errors of the estimated systolic BP and diastolic BP within the range of -0.19 +/- 3.3 mmHg and -1.29 +/- 2.6 mmHg, respectively. The mean absolute errors for systolic BP and diastolic BP are 3.2 mmHg and 2.6 mmHg, respectively. The accuracy of BPs estimated from the proposed method satisfies the requirements of the IEEE standard of 5 +/- 8 mmHg deviation, and thus, it may be used for ubiquitous long term blood pressure monitoring.

In this work, a seismocardiogram (SCG) based breathing-state measuring method is proposed for m-health applications. The aim of the proposed framework is to assess the human respiratory system by identifying degree-of-breathings, such as breathlessness, normal breathing, and long and labored breathing. For this, it is needed to measure cardiac-induced chest-wall vibrations, reflected in the SCG signal. Orthogonal subspace projection is employed to extract the SCG cycles with the help of a concurrent ECG signal. Subsequently, fifteen statistically significant morphological-features are extracted from each of the SCG cycles. These features can efficiently characterize physiological changes due to varying respiratory rates. Stacked autoencoder (SAE) based architecture is employed for the identification of different respiratory-effort levels. The performance of the proposed method is evaluated and compared with other standard classifiers for 1147 analyzed SCG-beats. The proposed method gives an overall average accuracy of 91.45% in recognizing three different breathing states. The quantitative analysis of the performance results clearly shows the effectiveness of the proposed framework. It may be employed in various healthcare applications, such as pre-screening medical sensors and IoT based remote health-monitoring systems.

Precise estimation of fiducial points of a seismocardiogram (SCG) signal is a challenging problem for its clinical usage. Delineation techniques proposed in the existing literature do not estimate all the clinically significant points of an SCG signal, simultaneously. The aim of this research work is to propose a delineation framework to identify IM, AO, IC, AC, pAC and MO fiducial points with the help of a PPG signal. The proposed delineation method processes a wavelet-based scalographic PPG and an envelope construction scheme is proposed to estimate the prominent peaks of the PPG signal. A set of amplitude histogram based decision rules is developed for estimation of SCG diastole phases, namely AC, pAC and MO. Subsequently, the systolic phases, IM, AO and IC are detected by applying diastole masking on SCG and decision rules. Experimental results on real-time SCG signals acquired from our designed data acquisition-circuitry and their analysis show the effectiveness of the proposed scheme. Additionally, these estimated parameters are analyzed to show the discrimination between normal breathing and breathlessness conditions., Comment: IEEE Transactions on Instrumentation and Measurement, 2020