A review of machine learning methods for non-invasive blood pressure estimation.

Blood pressure is a very important clinical measurement, offering valuable insights into the hemodynamic status of patients. Regular monitoring is crucial for early detection, prevention, and treatment of conditions like hypotension and hypertension, both of which increasing morbidity for a wide variety of reasons. This monitoring can be done either invasively or non-invasively and intermittently vs. continuously. An invasive method is considered the gold standard and provides continuous measurement, but it carries higher risks of complications such as infection, bleeding, and thrombosis. Non-invasive techniques, in contrast, reduce these risks and can provide intermittent or continuous blood pressure readings. This review explores modern machine learning-based non-invasive methods for blood pressure estimation, discussing their advantages, limitations, and clinical relevance.

Development and Validation of a Deep Learning Model for Prediction of Adult Physiological Deterioration.

BACKGROUND: Prediction-based strategies for physiologic deterioration offer the potential for earlier clinical interventions that improve patient outcomes. Current strategies are limited because they operate on inconsistent definitions of deterioration, attempt to dichotomize a dynamic and progressive phenomenon, and offer poor performance. OBJECTIVE: Can a deep learning deterioration prediction model (Deep Learning Enhanced Triage and Emergency Response for Inpatient Optimization [DETERIO]) based on a consensus definition of deterioration (the Adult Inpatient Decompensation Event [AIDE] criteria) and that approaches deterioration as a state "value-estimation" problem outperform a commercially available deterioration score? DERIVATION COHORT: The derivation cohort contained retrospective patient data collected from both inpatient services (inpatient) and emergency departments (EDs) of two hospitals within the University of California San Diego Health System. There were 330,729 total patients; 71,735 were inpatient and 258,994 were ED. Of these data, 20% were randomly sampled as a retrospective "testing set." VALIDATION COHORT: The validation cohort contained temporal patient data. There were 65,898 total patients; 13,750 were inpatient and 52,148 were ED. PREDICTION MODEL: DETERIO was developed and validated on these data, using the AIDE criteria to generate a composite score. DETERIO's architecture builds upon previous work. DETERIO's prediction performance up to 12 hours before T0 was compared against Epic Deterioration Index (EDI). RESULTS: In the retrospective testing set, DETERIO's area under the receiver operating characteristic curve (AUC) was 0.797 and 0.874 for inpatient and ED subsets, respectively. In the temporal validation cohort, the corresponding AUC were 0.775 and 0.856, respectively. DETERIO outperformed EDI in the inpatient validation cohort (AUC, 0.775 vs. 0.721; p < 0.01) while maintaining superior sensitivity and a comparable rate of false alarms (sensitivity, 45.50% vs. 30.00%; positive predictive value, 20.50% vs. 16.11%). CONCLUSIONS: DETERIO demonstrates promise in the viability of a state value-estimation approach for predicting adult physiologic deterioration. It may outperform EDI while offering additional clinical utility in triage and clinician interaction with prediction confidence and explanations. Additional studies are needed to assess generalizability and real-world clinical impact.

Making the Improbable Possible: Generalizing Models Designed for a Syndrome-Based, Heterogeneous Patient Landscape.

Syndromic conditions, such as sepsis, are commonly encountered in the intensive care unit. Although these conditions are easy for clinicians to grasp, these conditions may limit the performance of machine-learning algorithms. Individual hospital practice patterns may limit external generalizability. Data missingness is another barrier to optimal algorithm performance and various strategies exist to mitigate this. Recent advances in data science, such as transfer learning, conformal prediction, and continual learning, may improve generalizability of machine-learning algorithms in critically ill patients. Randomized trials with these approaches are indicated to demonstrate improvements in patient-centered outcomes at this point.

Applying Graph Theory to Examine the Dynamics of Student Discussions in Small-Group Learning.

Group work in science, technology, engineering, and mathematics courses is an effective means of improving student outcomes, and many different factors can influence the dynamics of student discussions and, ultimately, the success of collaboration. The substance and dynamics of group discussions are commonly examined using qualitative methods such as discourse analysis. To complement existing work in the literature, we developed a quantitative methodology that uses graph theory to map the progression of talk-turns of discussions within a group. We observed groups of students working with peer facilitators to solve problems in biological sciences, with three iterations of data collection and two major refinements of graph theory calculations. Results include general behaviors based on the turns in which different individuals talk and graph theory parameters to quantify group characteristics. To demonstrate the potential utility of the methodology, we present case studies with distinct patterns: a centralized group in which the peer facilitator behaves like an authority figure, a decentralized group in which most students talk their fair share of turns, and a larger group with subgroups that have implications for equity, diversity, and inclusion. Together, these results demonstrate that our adaptation of graph theory is a viable quantitative methodology to examine group discussions.

Muscarinic M1 receptor and cannabinoid CB1 receptor do not modulate paraoxon-induced seizures.

One of the major signs of severe organophosphate poisoning is seizures. Previous studies have shown that both muscarinic agonist- and organophosphate-induced seizures require activation of muscarinic acetylcholine receptors in the central nervous system. Seizures induced by the muscarinic agonist pilocarpine require the M1 receptor and are modulated by cannabinoid CB1 receptors. In this study, we determined whether M1 and CB1 receptors also regulated seizures induced by the organophosphate paraoxon. We found no differences in seizures induced by paraoxon in wild-type (WT) and M1 knockout (KO) mice, indicating that in contrast to pilocarpine seizures, M1 receptors are not required for paraoxon seizures. Furthermore, we found that pilocarpine administration resulted in seizure-independent activation of ERK in the hippocampus in a M1 receptor-dependent manner, while paraoxon did not induce seizure-independent activation of ERK in the mouse hippocampus. This shows that pilocarpine and paraoxon activated M1 receptors in the hippocampus to different extents. There were no differences in seizures induced by paraoxon in WT and CB1 KO mice, and neither CB1 agonist nor antagonist administration had significant effects on paraoxon seizures, indicating that, in contrast to pilocarpine seizures, paraoxon seizures are not modulated by CB1 receptors. These results demonstrate that there are fundamental molecular differences in the regulation of seizures induced by pilocarpine and paraoxon.

Modulation of pilocarpine-induced seizures by cannabinoid receptor 1.

Administration of the muscarinic agonist pilocarpine is commonly used to induce seizures in rodents for the study of epilepsy. Activation of muscarinic receptors has been previously shown to increase the production of endocannabinoids in the brain. Endocannabinoids act at the cannabinoid CB1 receptors to reduce neurotransmitter release and the severity of seizures in several models of epilepsy. In this study, we determined the effect of CB1 receptor activity on the induction in mice of seizures by pilocarpine. We found that decreased activation of the CB1 receptor, either through genetic deletion of the receptor or treatment with a CB1 antagonist, increased pilocarpine seizure severity without modifying seizure-induced cell proliferation and cell death. These results indicate that endocannabinoids act at the CB1 receptor to modulate the severity of pilocarpine-induced seizures. Administration of a CB1 agonist produced characteristic CB1-dependent behavioral responses, but did not affect pilocarpine seizure severity. A possible explanation for the lack of effect of CB1 agonist administration on pilocarpine seizures, despite the effects of CB1 antagonist administration and CB1 gene deletion, is that muscarinic receptor-stimulated endocannabinoid production is acting maximally at CB1 receptors to modulate sensitivity to pilocarpine seizures.

RACK1 associates with muscarinic receptors and regulates M(2) receptor trafficking.

Receptor internalization from the cell surface occurs through several mechanisms. Some of these mechanisms, such as clathrin coated pits, are well understood. The M(2) muscarinic acetylcholine receptor undergoes internalization via a poorly-defined clathrin-independent mechanism. We used isotope coded affinity tagging and mass spectrometry to identify the scaffolding protein, receptor for activated C kinase (RACK1) as a protein enriched in M(2)-immunoprecipitates from M(2)-expressing cells over those of non-M(2) expressing cells. Treatment of cells with the agonist carbachol disrupted the interaction of RACK1 with M(2). We further found that RACK1 overexpression inhibits the internalization and subsequent down regulation of the M(2) receptor in a receptor subtype-specific manner. Decreased RACK1 expression increases the rate of agonist internalization of the M(2) receptor, but decreases the extent of subsequent down-regulation. These results suggest that RACK1 may both interfere with agonist-induced sequestration and be required for subsequent targeting of internalized M(2) receptors to the degradative pathway.