Critical Transitions in Intensive Care Units: A Sepsis Case Study

Time-series-analysis-based detection of critical transitions in real-world non-autonomous systems

Real-world non-autonomous systems are open, out-of-equilibrium systems that evolve in and are driven by temporally varying environments. Such systems can show multiple timescale and transient dynamics together with transitions to very different and, at times, even disastrous dynamical regimes. Since such critical transitions disrupt the systems' intended or desired functionality, it is crucial to understand the underlying mechanisms, to identify precursors of such transitions, and to reliably detect them in time series of suitable system observables to enable forecasts. This review critically assesses the various steps of investigation involved in time-series-analysis-based detection of critical transitions in real-world non-autonomous systems: from the data recording to evaluating the reliability of offline and online detections. It will highlight pros and cons to stimulate further developments, which would be necessary to advance understanding and forecasting nonlinear behavior such as critical transitions in complex systems.

Application of early warning signs to physiological contexts: a comparison of multivariate indices in patients on long-term hemodialysis

Early warnings signs (EWSs) can anticipate abrupt changes in system state, known as "critical transitions," by detecting dynamic variations, including increases in variance, autocorrelation (AC), and cross-correlation. Numerous EWSs have been proposed; yet no consensus on which perform best exists. Here, we compared 15 multivariate EWSs in time series of 763 hemodialyzed patients, previously shown to present relevant critical transition dynamics. We calculated five EWSs based on AC, six on variance, one on cross-correlation, and three on AC and variance. We assessed their pairwise correlations, trends before death, and mortality predictive power, alone and in combination. Variance-based EWSs showed stronger correlations (r = 0.663 ± 0.222 vs. 0.170 ± 0.205 for AC-based indices) and a steeper increase before death. Two variance-based EWSs yielded HR95 > 9 (HR95 standing for a scale-invariant metric of hazard ratio), but combining them did not improve the area under the receiver-operating curve (AUC) much compared to using them alone (AUC = 0.798 vs. 0.796 and 0.791). Nevertheless, the AUC reached 0.825 when combining 13 indices. While some indicators did not perform overly well alone, their addition to the best performing EWSs increased the predictive power, suggesting that indices combination captures a broader range of dynamic changes occurring within the system. It is unclear whether this added benefit reflects measurement error of a unified phenomenon or heterogeneity in the nature of signals preceding critical transitions. Finally, the modest predictive performance and weak correlations among some indices call into question their validity, at least in this context.

Artificial intelligence for clinical decision support for monitoring patients in cardiovascular ICUs: A systematic review

Background

Artificial intelligence (AI) and machine learning (ML) models continue to evolve the clinical decision support systems (CDSS). However, challenges arise when it comes to the integration of AI/ML into clinical scenarios. In this systematic review, we followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA), the population, intervention, comparator, outcome, and study design (PICOS), and the medical AI life cycle guidelines to investigate studies and tools which address AI/ML-based approaches towards clinical decision support (CDS) for monitoring cardiovascular patients in intensive care units (ICUs). We further discuss recent advances, pitfalls, and future perspectives towards effective integration of AI into routine practices as were identified and elaborated over an extensive selection process for state-of-the-art manuscripts.

Methods

Studies with available English full text from PubMed and Google Scholar in the period from January 2018 to August 2022 were considered. The manuscripts were fetched through a combination of the search keywords including AI, ML, reinforcement learning (RL), deep learning, clinical decision support, and cardiovascular critical care and patients monitoring. The manuscripts were analyzed and filtered based on qualitative and quantitative criteria such as target population, proper study design, cross-validation, and risk of bias.

Results

More than 100 queries over two medical search engines and subjective literature research were developed which identified 89 studies. After extensive assessments of the studies both technically and medically, 21 studies were selected for the final qualitative assessment.

Discussion

Clinical time series and electronic health records (EHR) data were the most common input modalities, while methods such as gradient boosting, recurrent neural networks (RNNs) and RL were mostly used for the analysis. Seventy-five percent of the selected papers lacked validation against external datasets highlighting the generalizability issue. Also, interpretability of the AI decisions was identified as a central issue towards effective integration of AI in healthcare.

Synchrony of biomarker variability indicates a critical transition: Application to mortality prediction in hemodialysis

Critical transition theory suggests that complex systems should experience increased temporal variability just before abrupt state changes. We tested this hypothesis in 763 patients on long-term hemodialysis, using 11 biomarkers collected every two weeks and all-cause mortality as a proxy for critical transitions. We find that variability-measured by coefficients of variation (CVs)-increases before death for all 11 clinical biomarkers, and is strikingly synchronized across all biomarkers: the first axis of a principal component analysis on all CVs explains 49% of the variance. This axis then generates powerful predictions of mortality (HR95 = 9.7, p < 0.0001, where HR95 is a scale-invariant metric of hazard ratio; AUC up to 0.82) and starts to increase markedly ∼3 months prior to death. Our results provide an early warning sign of physiological collapse and, more broadly, a quantification of joint system dynamics that opens questions of how system modularity may break down before critical transitions.

Ground truth labels challenge the validity of sepsis consensus definitions in critical illness

BACKGROUND

Sepsis is the leading cause of death in the intensive care unit (ICU). Expediting its diagnosis, largely determined by clinical assessment, improves survival. Predictive and explanatory modelling of sepsis in the critically ill commonly bases both outcome definition and predictions on clinical criteria for consensus definitions of sepsis, leading to circularity. As a remedy, we collected ground truth labels for sepsis.

METHODS

In the Ground Truth for Sepsis Questionnaire (GTSQ), senior attending physicians in the ICU documented daily their opinion on each patient's condition regarding sepsis as a five-category working diagnosis and nine related items. Working diagnosis groups were described and compared and their SOFA-scores analyzed with a generalized linear mixed model. Agreement and discriminatory performance measures for clinical criteria of sepsis and GTSQ labels as reference class were derived.

RESULTS

We analyzed 7291 questionnaires and 761 complete encounters from the first survey year. Editing rates for all items were > 90%, and responses were consistent with current understanding of critical illness pathophysiology, including sepsis pathogenesis. Interrater agreement for presence and absence of sepsis was almost perfect but only slight for suspected infection. ICU mortality was 19.5% in encounters with SIRS as the "worst" working diagnosis compared to 5.9% with sepsis and 5.9% with severe sepsis without differences in admission and maximum SOFA. Compared to sepsis, proportions of GTSQs with SIRS plus acute organ dysfunction were equal and macrocirculatory abnormalities higher (p < 0.0001). SIRS proportionally ranked above sepsis in daily assessment of illness severity (p < 0.0001). Separate analyses of neurosurgical referrals revealed similar differences. Discriminatory performance of Sepsis-1/2 and Sepsis-3 compared to GTSQ labels was similar with sensitivities around 70% and specificities 92%. Essentially no difference between the prevalence of SIRS and SOFA ≥ 2 yielded sensitivities and specificities for detecting sepsis onset close to 55% and 83%, respectively.

CONCLUSIONS

GTSQ labels are a valid measure of sepsis in the ICU. They reveal suspicion of infection as an unclear clinical concept and refute an illness severity hierarchy in the SIRS-sepsis-severe sepsis spectrum. Ground truth challenges the accuracy of Sepsis-1/2 and Sepsis-3 in detecting sepsis onset. It is an indispensable intermediate step towards advancing diagnosis and therapy in the ICU and, potentially, other health care settings.

A Locally Optimized Data-Driven Tool to Predict Sepsis-Associated Vasopressor Use in the ICU

OBJECTIVES

To train a model to predict vasopressor use in ICU patients with sepsis and optimize external performance across hospital systems using domain adaptation, a transfer learning approach.

DESIGN

Observational cohort study.

SETTING

Two academic medical centers from January 2014 to June 2017.

PATIENTS

Data were analyzed from 14,512 patients (9,423 at the development site and 5,089 at the validation site) who were admitted to an ICU and met Center for Medicare and Medicaid Services definition of severe sepsis either before or during the ICU stay. Patients were excluded if they never developed sepsis, if the ICU length of stay was less than 8 hours or more than 20 days or if they developed shock up to the first 4 hours of ICU admission.

MEASUREMENTS AND MAIN RESULTS

Forty retrospectively collected features from the electronic medical records of adult ICU patients at the development site (four hospitals) were used as inputs for a neural network Weibull-Cox survival model to derive a prediction tool for future need of vasopressors. Domain adaptation updated parameters to optimize model performance in the validation site (two hospitals), a different healthcare system over 2,000 miles away. The cohorts at both sites were randomly split into training and testing sets (80% and 20%, respectively). When applied to the test set in the development site, the model predicted vasopressor use 4-24 hours in advance with an area under the receiver operator characteristic curve, specificity, and positive predictive value ranging from 0.80 to 0.81, 56.2% to 61.8%, and 5.6% to 12.1%, respectively. Domain adaptation improved performance of the model to predict vasopressor use within 4 hours at the validation site (area under the receiver operator characteristic curve 0.81 [CI, 0.80-0.81] from 0.77 [CI, 0.76-0.77], p < 0.01; specificity 59.7% [CI, 58.9-62.5%] from 49.9% [CI, 49.5-50.7%], p < 0.01; positive predictive value 8.9% [CI, 8.5-9.4%] from 7.3 [7.1-7.4%], p < 0.01).

CONCLUSIONS

Domain adaptation improved performance of a model predicting sepsis-associated vasopressor use during external validation.

Prediction of Mortality in Hemodialysis Patients Using Moving Multivariate Distance

There is an increasingly widespread use of biomarkers in network physiology to evaluate an organism’s physiological state. A recent study showed that albumin variability increases before death in chronic hemodialysis patients. We hypothesized that a multivariate statistical approach would better allow us to capture signals of impending physiological collapse/death. We proposed a Moving Multivariate Distance (MMD), based on the Mahalanobis distance, to quantify the variability of the multivariate biomarker profile as a whole from one visit to the next. Biomarker profiles from a visit were used as the reference to calculate MMD at the subsequent visit. We selected 16 biomarkers (of which 11 are measured every 2 weeks) from blood samples of 763 chronic kidney disease patients hemodialyzed at the CHUS hospital in Quebec, who visited the hospital regularly (∼every 2 weeks) to perform routine blood tests. MMD tended to increase markedly preceding death, indicating an increasing intraindividual multivariate variability presaging a critical transition. In survival analysis, the hazard ratio between the 97.5th percentile and the 2.5th percentile of MMD reached as high as 21.1 [95% CI: 14.3, 31.2], showing that higher variability indicates substantially higher mortality risk. Multivariate approaches to early warning signs of critical transitions hold substantial clinical promise to identify early signs of critical transitions, such as risk of death in hemodialysis patients; future work should also explore whether the MMD approach works in other complex systems (i.e., ecosystems, economies), and should compare it to other multivariate approaches to quantify system variability.

WISDoM: Characterizing Neurological Time Series With the Wishart Distribution

WISDoM (Wishart Distributed Matrices) is a framework for the quantification of deviation of symmetric positive-definite matrices associated with experimental samples, such as covariance or correlation matrices, from expected ones governed by the Wishart distribution. WISDoM can be applied to tasks of supervised learning, like classification, in particular when such matrices are generated by data of different dimensionality (e.g., time series with same number of variables but different time sampling). We show the application of the method in two different scenarios. The first is the ranking of features associated with electro encephalogram (EEG) data with a time series design, providing a theoretically sound approach for this type of studies. The second is the classification of autistic subjects of the Autism Brain Imaging Data Exchange study using brain connectivity measurements.