Identification of physiological shock in intensive care units via Bayesian regime switching models

Abstract

Detection of occult hemorrhage (i.e., internal bleeding) in patients in intensive care units (ICUs) can pose significant challenges for critical care workers. Because blood loss may not always be clinically apparent, clinicians rely on monitoring vital signs for specific trends indicative of a hemorrhage event. The inherent difficulties of diagnosing such an event can lead to late intervention by clinicians which has catastrophic consequences. Therefore, a methodology for early detection of hemorrhage has wide utility. We develop a Bayesian regime switching model (RSM) that analyzes trends in patients' vitals and labs to provide a probabilistic assessment of the underlying physiological state that a patient is in at any given time. This article is motivated by a comprehensive dataset we curated from Mayo Clinic of 33,924 real ICU patient encounters. Longitudinal response measurements are modeled as a vector autoregressive process conditional on all latent states up to the current time point, and the latent states follow a Markov process. We present a novel Bayesian sampling routine to learn the posterior probability distribution of the latent physiological states, as well as develop an approach to account for pre-ICU-admission physiological changes. A simulation and real case study illustrate the effectiveness of our approach.

Figures (10)

• Figure 1: Schematic of the model dependence structure for an HMM, an AR-HMM of order one, and the RSM used in our approach, respectively, from left to right. Let $\vb*{y}_k$ be some observed response vector at a time instance $k$ and $s_k$ be the corresponding latent state.
• Figure 1: Box plots of the posterior means across 25 simulated datasets. These four columns specifically compare the estimated slope coefficients corresponding to state 2 (columns 1 and 2) and state 3 (columns 3 and 4) using Model A and Model B to fit the data. The true parameter value is provided on the x-axis, as well as marked by a red horizontal line, in each plot.
• Figure 2: All allowable transitions for the five physiological states.
• Figure 2: The top two panels correspond to the longitudinal vital measurements. The points with the error bars correspond to missing values; the error bars are empirical 95% credible intervals for the imputed response values. The third panel depicts the discrete posterior probability distributions of the latent states, at each time point. The bottom panel is the posterior probability of state 2 at each time point, and the yellow dashed line represents the threshold $\hat{c} = 0.0465$ determined from Section \ref{['chap3:subsec:postInt']}. The white stars indicate that the posterior probability of state 2 exceeds the threshold. The purple and turquoise vertical lines represent RBC transfusion order and administration times, respectively.
• Figure 3: ROC curve depicting the sensitivity and specificity of identifying state 2. The red circle indicates the posterior probability threshold $c \in [0,1]$ that minimizes FPR - TPR.