Pressure reactivity index (PRx) is a surrogate for cerebral autoregulation and has been used for prognostication in aneurysmal subarachnoid hemorrhage (SAH). We examined patient-specific temporal courses of PRx and identified time thresholds that optimized the accuracy of PRx monitoring. Comatose patients with SAH from two Comprehensive Stroke Centers were identified and received continuous ICP and PRx recordings. Outcomes were dichotomized into “poor” versus “good” based on disposition and modified Rankin score. Smoothed PRx trajectories were created to generate “candidate features”, looking at daily average PRx and cumulative first-order and second-order changes in PRx. “Candidate features” were used to perform penalized logistic regression analysis. Penalized logistic regression models that maximized specificity for poor outcome were iteratively generated and evaluated sensitivity changes over time. We evaluated 33 comatose SAH patients in this cohort. Average PRx trajectories for good and poor outcome groups diverged at post-ictus day 6. When targeting specificities ≥ 78.6

BACKGROUND:Transcranial Doppler mean flow velocity (MFV) and pulsatility index (PI) are used for ancillary monitoring in aneurysmal subarachnoid hemorrhage (SAH). We evaluated PI and MFV variables in the middle cerebral arteries (MCA) for potential associations with delayed cerebral ischemia (DCI) and clinical outcome. METHODS:We retrospectively evaluated patients with SAH over a six-year period. Poor outcome was defined as a three-month Modified Rankin Scale (Rajajee et al., 2012; Bellner et al., n.d.; Chang et al., 2023), and DCI was defined by presence of vascular infarcts on brain imaging. Serial PI and MFV values for each patient were compiled to obtain four PI and MFV variables, including maximum absolute change in daily PI values (AbsΔPI). Multivariate logistic regression was performed to identify subsets of variables that maximized area under curve-receiver operating characteristic (AUC-ROC) for clinical outcome. Multivariate logistic regression analysis using clinical outcome and DCI was generated using PI variables, MFV variables, and demographic variables. RESULTS:268 patients met inclusion criteria and were evaluated. PI, MFV, and demographic variables yielded an AUC-ROC value of 0.84 for clinical outcome. Multivariate logistic regression analysis showed that AbsΔPI (p = 0.01), older age (p < 0.001), Hunt Hess score (p < 0.001), and Fisher score (p = 0.05) were significant predictors for poor clinical outcome. No MFV variable had significant associations with clinical outcome. CONCLUSION:We found that PI variability in the MCAs had a significant association with clinical outcome in patients with SAH: every 0.1 variability in PI resulted in 1.4-times higher odds of poor clinical outcome. Further studies are needed for confirmation.

The exquisite eggshell mimicry found in many cuckoo-host systems has been the textbook example of a coevolutionary arms race 1 . In this system, cuckoos lay their eggs in their hosts’ nests leaving these foster parents to rear their young 2 . In defence, most of these hosts have evolved the ability to recognize cuckoo eggs. It is long-assumed that these interactions select for ever-improving eggshell mimicry in cuckoo host-races laying distinct eggs that specifically mimic their host’s eggshell, regardless of its particular colour 3,4 . We challenge this paradigm and demonstrate that coevolutionary arms races and egg colour mimicry are the exception to the rule. Here, using eggshell reflectance spectra from 39 host species and the specific cuckoo eggs found in their nests, we show that eggshell colour mimicry is very uncommon even among frequent hosts. Although cuckoo eggs display considerable variation in eggshell colour, their hosts’ eggshell colours occupy 6 times their colour volume. As a result, mimicry is only achieved for a small subset of the egg colour gamut. Furthermore, cuckoo host-races rarely lay eggs that match their hosts’ eggs better than those of other potential hosts or that are distinct from those laid by other cuckoo host-races. These patterns are inconsistent with the long-held assumption of coevolutionary arms races, but consistent with coevolutionary alternation that is defined by a process of continual host-switching. These findings open new avenues for research in a system that has long served as a model for studying coevolutionary processes.

Robust estimators for linear regression require non-convex objective functions to shield against adverse effects of contamination, including outliers. This non-convexity brings challenges, particularly when combined with penalization in high-dimensional settings. A crucial challenge is selecting hyperparameters for the penalty based on a finite sample. In practice, cross-validation (CV) is the prevalent strategy with good performance for convex estimators. Applied with robust estimators, however, CV often gives subpar results due to the interplay between multiple local minima and the penalty. The best local minimum attained on the full training data may not be the minimum with the desired statistical properties. Furthermore, there may be a mismatch between this minimum and the minima attained in the CV folds which are used for evaluating the prediction error. This article introduces a novel adaptive CV strategy that tracks multiple minima for each combination of hyperparameters and subsets of the data. A matching scheme is presented for correctly evaluating minima computed on the full training data using the best-matching minima from the CV folds. We show that the proposed strategy reduces the variability of the estimated performance metric, leads to smoother CV curves, and therefore substantially increases the reliability and utility of robust penalized estimators. Supplementary materials for this article are available online.

Purpose: Cerebral autoregulation monitoring utilizes pressure reactivity index (PRx) but has limited use due to its reliance on invasive continuous intracranial pressure monitoring. Non-invasive forms of cerebral autoregulation monitoring may help evaluate secondary neurological injury. We compared temporal trends for tissue oxygen reactivity index (TOx)—a non-invasive cerebral autoregulation index that utilizes the moving correlation coefficient between tissue oxygen saturation and mean arterial pressure—and PRx. Methods: A single-center, retrospective exploratory feasibility study in a Level-One Trauma and Tertiary Stroke center comparing PRx and TOx smoothed trajectories in patients with severe neurological injury and comatose exams. Results: Eighteen patients who received TOx and PRx monitoring were analyzed. Individualized smoothed trajectories for TOx and PRx were created. Overall median correlations between TOx and PRx was 0.73. TOx and PRx were also analyzed for their association with poor clinical outcome. Matthew Correlation Coefficient (MCC) suggested that TOx had the strongest association with outcome, reaching a peak value of 0.71 that occurred after day 13 of monitoring. PRx achieved a peak MCC of 0.6 that was consistently reached after day 15 of monitoring. Conclusion: Our findings suggest that TOx may correlate with PRx trajectories. In our cohort, TOx was more strongly associated with poor outcome than PRx. These preliminary results should be interpreted as exploratory, and further research is required to confirm and validate these findings.

Guideline-directed medical therapy utilization in patients with HFrEF remains low despite benefits in morbidity and mortality. The authors describe a unique Quality Improvement initiative designed to increase ARNI and MRA utilization in outpatients with HFrEF in a large cardiology practice, whereby eligible patients were identified in a standardized review process and medication utilization rates were linked to group quality metrics. Eligible HFrEF patients were defined as having an LVEF ≤ 40% and NYHA Class II-IV level of symptoms. Those with an LVEF > 40%, no documented LVEF, or with NYHA Class I symptoms were excluded. ARNI utilization was defined as any dose of sacubitril/valsartan prescribed, and MRA utilization was defined as any dose of either spironolactone or eplerenone prescribed. Group quality metric targets were set at > 25% ARNI prescription and > 60% MRA prescription in eligible patients. Following project implementation, ARNI utilization rose from 31% to 67% and MRA increased from 28% to 66%. Establishing clear quality metrics and formulating a proactive evaluation process was associated with a significant increase in prescription rates.

INTRODUCTION:Acute respiratory distress syndrome (ARDS) patients are at risk of thrombosis through mechanisms implicating oxidized low-density lipoprotein (oxLDL). Endothelial cells, immune cells and platelets were reported to express scavenger receptors for oxLDL: Lox-1 and CD36. We hypothesized that platelets shed a soluble Lox-1 ectodomain (sLox-1) and release CD36-bearing procoagulant microparticles (MPs), that both become elevated in subjects with ARDS-induced coagulopathy. METHODS:Using anti-extracellular and anti-intracellular Lox-1 antibodies, we first tested by western blot whether platelets express Lox-1 and shed sLox-1 upon activation. Next, we measured sLox-1 in blood plasma of 23 healthy donors and 48 ARDS Omega patients with and without coagulopathy, and assessed the corresponding MP fraction for Lox-1/sLox-1 and CD36. We evaluated mechanisms of sLox-1-MP association. Recombinant proteins were used as controls. RESULTS:Resting platelets expressed abundant CD36 (7.8 ng/μg protein extract) which was released upon oxLDL stimulation, but undetectable levels of full-length 37 kDa Lox-1 receptor or 24 kDa sLox-1 (below 10 pg/μg). In an RNAseq meta-analysis, platelets expressed negligible OLR1, the mRNA encoding Lox-1, compared to CD36. A subset of ARDS patients showed elevated plasma sLox-1 and MP-associated sLox-1 compared to healthy controls that was positively associated with 90-day survival and low coagulopathy. MP-associated CD36 was reduced in ARDS plasma compared to healthy donors and did not correlate with survival, coagulopathy, or sLox-1. oxLDL promoted sLox-1 binding to CD36-deficient MPs. CONCLUSION:sLox-1 arising from a non-platelet cell source associates with circulating MPs which could serve a protective role in ARDS.

BACKGROUND AND OBJECTIVES:Severe traumatic brain injury (sTBI) represents a diffuse, heterogeneous disease where therapeutic targets for optimizing clinical outcome remain unclear. Mean pressure reactivity index (PRx) values have demonstrated associations with clinical outcome in sTBI. However, the retrospective derivation of a mean value diminishes its bedside significance. We evaluated PRx temporal profiles for patients with sTBI and identified time thresholds suggesting optimal neuroprognostication. METHODS:Patients with sTBI and continuous bolt intracranial pressure monitoring were identified. Outcomes were dichotomized by disposition status ("good outcome" was denoted by home and acute rehabilitation). PRx values were obtained every minute by taking moving correlation coefficients of intracranial pressures and mean arterial pressures. Average PRx trajectories for good and poor outcome groups were calculated by extending the last daily averaged PRx value to day 18. Each patient also had smoothed PRx trajectories that were used to generate "candidate features." These "candidate features" included daily average PRx's, cumulative first-order changes in PRx and cumulative second-order changes in PRx. Changes in sensitivity over time for predicting poor outcome was then evaluated by generating penalized logistic regression models that were derived from the "candidate features" and maximized specificity. RESULTS:Among 33 patients with sTBI, 18 patients achieved good outcome and 15 patients had poor outcome. Average PRx trajectories for the good and poor outcome groups started on day 6 and consistently diverged at day 9. When targeting a specificity >83.3%, an 85% maximum sensitivity for determining poor outcome was achieved at hospital day 6. Subsequent days of PRx monitoring showed diminishing sensitivities. CONCLUSION:Our findings suggest that in a population of sTBI, PRx sensitivities for predicting poor outcome was maximized at hospital day 6. Additional study is warranted to validate this model in larger populations.

Background Distal transradial access (dTRA) is an alternative to conventional forearm transradial access (fTRA) for coronary angiography (CAG). Differences in healing of the radial artery (RA) in the forearm have not been evaluated between these 2 access strategies. We sought to compare the mean difference in forearm RA intimal‐medial thickening (IMT) in patients randomized to dTRA versus fTRA. Methods and Results In this single‐center randomized clinical trial, 64 patients undergoing nonemergent CAG were randomized (1:1) to dTRA versus fTRA. Ultra–high‐resolution (55‐MHz) vascular ultrasound of the forearm and distal RA was performed pre‐CAG and at 90 days. The primary end point was the mean change in forearm RA IMT. Secondary end points included procedural characteristics, vascular injury, RA occlusion, and ipsilateral hand pain and function. Baseline demographics and clinical characteristics, mean forearm RA IMT, and procedural specifics were similar between the dTRA and fTRA cohorts. There was no difference in mean change in forearm RA IMT between the 2 cohorts (0.07 versus 0.07 mm; P=0.37). No RA occlusions or signs of major vascular injury were observed at 90 days. Ipsilateral hand pain and function (Borg pain scale score: 12 versus 11; P=0.24; Disabilities of the Arm, Shoulders, and Hand scale score: 6 versus 8; P=0.46) were comparable. Conclusions Following CAG, dTRA was associated with no differences in mean change of forearm RA IMT, hand pain, and function versus fTRA for CAG. Further investigation is warranted to elucidate mechanisms and predictors of RA healing and identify effective strategies to preserving RA integrity for repeated procedures. Registration URL: https://www.clinicaltrials.gov; Unique identifier: NCT04801901.

Plants, animals, and fungi display a rich tapestry of colors. Animals, in particular, use colors in dynamic displays performed in spatially complex environments. Although current approaches for studying colors are objective and repeatable, they miss the temporal variation of color signals entirely. Here, we introduce hardware and software that provide ecologists and filmmakers the ability to accurately record animal-perceived colors in motion. Specifically, our Python codes transform photos or videos into perceivable units (quantum catches) for animals of known photoreceptor sensitivity. The plans and codes necessary for end-users to capture animal-view videos are all open source and publicly available to encourage continual community development. The camera system and the associated software package will allow ecologists to investigate how animals use colors in dynamic behavioral displays, the ways natural illumination alters perceived colors, and other questions that remained unaddressed until now due to a lack of suitable tools. Finally, it provides scientists and filmmakers with a new, empirically grounded approach for depicting the perceptual worlds of nonhuman animals.

Abstract Jonathan Auerbach, David Kepplinger and Nicholas Rios use two popular data science algorithms – naïve Bayes and eigencentrality – to examine the difference between data scientists, statisticians, and other occupations

INTRODUCTION: The pressure reactivity index (PRx) is a surrogate for vascular cerebral autoregulation that evaluates moving correlation coefficients between intracranial pressure (ICP) and mean arterial blood pressure (MAP). Ideal threshold PRx values for distinguishing between poor and good outcome in severe traumatic brain injury (sTBI) is unclear. METHODS: Retrospective evaluation of consecutive patients over a three-year period with sTBI and polytrauma who received continuous ICP, MAP, and PRx monitoring. Good outcome was defined as patients who upon hospital discharge had a modified Rankin score < 4 and were discharged to an acute rehabilitation facility or home. RESULTS: 36 patients received continuous monitoring. Nine patients were excluded for the following reasons: early palliative withdrawal, early death through multi-organ dysfunction not related to the sTBI, and continuous neuromonitoring < 24 hrs. Of the 27 patients evaluated, average PRx values for the good outcome group was 0.18 (+0.17) and 0.37 (+0.24) for the poor outcome group. Receiver operating characteristics (ROC) analysis showed that a PRx threshold value > 0.34 had the highest combined sensitivity and specificity for predicting poor outcome with an AUC (95% CI) = 0.775 (0.583-0.967). When dichotomized using ROC curve analysis, a threshold PRx value of greater than 0.34 was independently associated with higher odds of poor outcome (OR 18.63, 95% CI 1.17-297.21, p = 0.038) after adjustment for potential confounders. CONCLUSIONS: Higher mean PRx is an independent predictor for poor outcome in patients with sTBI. Further study is required to evaluate the temporal role of PRx changes to improve neuroprognostication.

Heavy-tailed error distributions and predictors with anomalous values are ubiquitous in high-dimensional regression problems and can seriously jeopardize the validity of statistical analyses if not properly addressed. For more reliable variable selection and prediction under these adverse conditions, adaptive PENSE, a new robust regularized regression estimator, is proposed. Adaptive PENSE yields reliable variable selection and coefficient estimates even under aberrant contamination in the predictors or residuals. It is shown that the adaptive penalty leads to more robust and reliable variable selection than other penalties, particularly in the presence of gross outliers in the predictor space. It is further demonstrated that adaptive PENSE has strong variable selection properties and that it possesses the oracle property even under heavy-tailed errors and without the need to estimate the error scale. Numerical studies on simulated and real data sets highlight the superior finite-sample performance in a vast range of settings compared to other robust regularized estimators in the case of contaminated samples. An R package implementing a fast algorithm for computing the proposed method and additional simulation results are provided in the supplementary materials.

Background: Pressure reactivity index (PRx) utilizes moving correlation coefficients from intracranial pressure (ICP) and mean arterial pressures to evaluate cerebral autoregulation. We evaluated patients with poor-grade subarachnoid hemorrhage (SAH), identified their PRx trajectories over time, and identified threshold time points where PRx could be used for neuroprognostication.Methods: Patients with poor-grade SAH were identified and received continuous bolt ICP measurements. Dichotomized outcomes were based on ninety-day modified Rankin scores and disposition. Smoothed PRx tra-jectories for each patient were created to generate "candidate features" that looked at daily average PRx, cu-mulative first-order changes in PRx, and cumulative second-order changes in PRx. "Candidate features" were then used to perform penalized logistic regression analysis using poor outcome as the dependent variable. Penalized logistic regression models that maximized specificity for poor outcome were generated over several time periods and evaluated how sensitivities changed over time.Results: 16 patients with poor-grade SAH were evaluated. Average PRx trajectories for the good (PRx < 0.25) and poor outcome groups (PRx > 0.5) started diverging at post-ictus day 8. When targeting specificities & GE;88% for poor outcome, sensitivities for poor outcome consistently increased to >70% starting at post-ictus days 12-14 with a maximum sensitivity of 75% occurring at day 18.Conclusions: Our results suggest that by using PRx trends, early neuroprognostication in patients with SAH and poor clinical exams may start becoming apparent at post-ictus day 8 and reach adequate sensitivities by post -ictus days 12-14. Further study is required to validate this in larger poor-grade SAH populations.

BACKGROUND:Blood clots are living tissues that release inflammatory mediators including IL-8/CXCL8 and MCP-1/CCL2. A deeper understanding of blood clots is needed to develop new therapies for prothrombotic disease states and regenerative medicine. OBJECTIVES:To identify a common transcriptional shift in cultured blood clot leukocytes. METHODS:Differential gene expression of whole blood and cultured clots (4 hours at 37 °C) was assessed by RNA sequencing (RNAseq), reverse transcriptase-polymerase chain reaction, proteomics, and histology (23 diverse healthy human donors). Cultured clot serum bioactivity was tested in endothelial barrier functional assays. RESULTS:All cultured clots developed a polymorphonuclear myeloid-derived suppressor cell (PMN-MDSC) signature, including up-regulation of OLR1 (mRNA encoding lectin-like oxidized low-density lipoprotein receptor 1 [Lox-1]), IL-8/CXCL8, CXCL2, CCL2, IL10, IL1A, SPP1, TREM1, and DUSP4/MKP. Lipopolysaccharide enhanced PMN-MDSC gene expression and specifically induced a type II interferon response with IL-6 production. Lox-1 was specifically expressed by cultured clot CD15+ neutrophils. Cultured clot neutrophils, but not activated platelets, shed copious amounts of soluble Lox-1 (sLox-1) with a donor-dependent amplitude. sLox-1 shedding was enhanced by phorbol ester and suppressed by heparin and by beta-glycerol phosphate, a phosphatase inhibitor. Cultured clot serum significantly enhanced endothelial cell monolayer barrier function, consistent with a proresolving bioactivity. CONCLUSION:This study suggests that PMN-MDSC activation is part of the innate immune response to coagulation which may have a protective role in inflammation. The cultured blood clot is an innovative thrombus model that can be used to study both sterile and nonsterile inflammatory states and could be used as a personalized medicine tool for drug screening.

Plants, animals, and fungi display a rich tapestry of colors. Animals, in particular, use colors in dynamic displays performed in spatially complex environments. In such natural settings, light is reflected or refracted from objects with complex shapes that cast shadows and generate highlights. In addition, the illuminating light changes continuously as viewers and targets move through heterogeneous, continually fluctuating, light conditions. Although traditional spectrophotometric approaches for studying colors are objective and repeatable, they fail to document this complexity. Worse, they miss the temporal variation of color signals entirely. Here, we introduce hardware and software that provide ecologists and filmmakers the ability to accurately record animal-perceived colors in motion. Specifically, our Python codes transform photos or videos into perceivable units (quantum catches) for any animal of known photoreceptor sensitivity. We provide the plans, codes, and validation tests necessary for end-users to capture animal-view videos. This approach will allow ecologists to investigate how animals use colors in dynamic behavioral displays, the ways natural illumination alters perceived colors, and other questions that remained unaddressed until now due to a lack of suitable tools. Finally, our pipeline provides scientists and filmmakers with a new, empirically grounded approach for depicting the perceptual worlds of non-human animals.

Adaptive PENSE is a method that can be used to build models for predicting clinical outcomes from a small subset of a potentially large number of candidate proteins. Adaptive PENSE is designed to give reliable results under two common challenges often encountered in these kinds of studies: (1) the number of samples with known clinical outcome and proteomic data is small, while the number of candidate proteins is large and/or (2) proteomic data and the clinical outcome measurements suffer from data quality issues in a small fraction of samples. Even in the presence of these challenges, adaptive PENSE reliably identifies proteins relevant for prediction and estimates accurate predictive models. Adaptive PENSE is designed to be resilient to data quality issues in up to 50% of samples. Almost half of the samples could have aberrant values in the measured protein levels and clinical outcome values without causing severe detrimental effects to the estimated predictive model. The method is implemented as an R package and supports the user in the model selection process by automating most steps and providing diagnostic visualizations to guide the user. Users can choose among several predictive models to select the model with high prediction accuracy and an appropriate number of selected proteins.

In large-scale quantitative proteomic studies, scientists measure the abundance of thousands of proteins from the human proteome in search of novel biomarkers for a given disease. Penalized regression estimators can be used to identify potential biomarkers among a large set of molecular features measured. Yet, the performance and statistical properties of these estimators depend on the loss and penalty functions used to define them. Motivated by a real plasma proteomic biomarkers study, we propose a new class of penalized robust estimators based on the elastic net penalty, which can be tuned to keep groups of correlated variables together in the selected model and maintain robustness against possible outliers. We also propose an efficient algorithm to compute our robust penalized estimators and derive a data-driven method to select the penalty term. Our robust penalized estimators have very good robustness properties and are also consistent under certain regularity conditions. Numerical results show that our robust estimators compare favorably to other robust penalized estimators. Using our proposed methodology for the analysis of the proteomics data, we identify new potentially relevant biomarkers of cardiac allograft vasculopathy that are not found with nonrobust alternatives. The selected model is validated in a new set of 52 test samples and achieves an area under the receiver operating characteristic (AUC) of 0.85.

In large-scale quantitative proteomic studies, scientists measure the abundance of hundreds or thousands of proteins from the human proteome in search of novel biomarkers for a given disease. Despite current innovations in biomedical technologies, advanced statistical and computational methods are still required to harness the rich information contained in these large and complex datasets. While penalized regression estimators can be used to identify potential biomarkers among a large set of molecular features, it is well-known that the performance and statistical properties of the selected model depend on the loss and penalty functions used to construct the regularized estimator. For example, the presence of outlying observations in the data can seriously affect classical estimators that penalize the square error loss function. Similarly, the choice of the penalty function in these estimators is important to be able to preserve groups of correlated proteins in the selected model. Thus, in this paper we propose a new class of penalized robust estimators based on the elastic net penalty, which can be tuned to keep groups of correlated variables together as they enter or leave the model, while protecting the resulting estimator against possibly aberrant observations in the dataset. Our robust penalized estimators have very good robustness properties and are also consistent under relatively weak assumptions. In this paper we also propose an efficient algorithm to compute our robust penalized estimators and we derive a data-driven method to select the penalty term, which is a critical part of any application with real data. Our numerical experiments show that our proposals compare favorably to other robust penalized estimators. Noteworthy, our robust estimators identify new potentially relevant biomarkers of cardiac allograft vasculopathy that are not found with non-robust alternatives. Importantly, the selected model is validated in a new set of 52 test samples, achieving an area under the receiver operating characteristic curve (AUC) of 0.85. *Supported by NSERC Discovery Grant MSC 2010 subject classifications: Primary 62J; secondary 62J05, 62J07, 62J07; Primary 62P; secondary 62P10; Primary 62F; secondary 62F35