Early Prediction of Cardiogenic Shock Using Machine Learning.

Cardiogenic shock (CS) is a severe condition with in-hospital mortality of up to 50%. Patients who develop CS may have previous cardiac history, but that may not always be the case, adding to the challenges in optimally identifying and managing these patients. Patients may present to a medical facility with CS or develop CS while in the emergency department (ED), in a general inpatient ward (WARD) or in the critical care unit (CC). While different clinical pathways for management exist once CS is recognized, there are challenges in identifying the patients in a timely manner, in all settings, in a timeframe that will allow proper management. We therefore developed and evaluated retrospectively a machine learning model based on the XGBoost (XGB) algorithm which runs automatically on patient data from the electronic health record (EHR). The algorithm was trained on 8 years of de-identified data (from 2010 to 2017) collected from a large regional healthcare system. The input variables include demographics, vital signs, laboratory values, some orders, and specific pre-existing diagnoses. The model was designed to make predictions 2 h prior to the need of first CS intervention (inotrope, vasopressor, or mechanical circulatory support). The algorithm achieves an overall area under curve (AUC) of 0.87 (0.81 in CC, 0.84 in ED, 0.97 in WARD), which is considered useful for clinical use. The algorithm can be refined based on specific elements defining patient subpopulations, for example presence of acute myocardial infarction (AMI) or congestive heart failure (CHF), further increasing its precision when a patient has these conditions. The top-contributing risk factors learned by the model are consistent with existing clinical findings. Our conclusion is that a useful machine learning model can be used to predict the development of CS. This manuscript describes the main steps of the development process and our results.

Neonatal ICU antibiotic use trends within an integrated delivery network.

BACKGROUND AND OBJECTIVES: There is a need for robust antibiotic stewardship programs (ASPs) in the neonatal population. This study's objectives were to assess neonatal antibiotic use practices over an extended period across an integrated delivery network (IDN), including six Neonatal Intensive Care Units (NICUs), to identify those most successful practices reducing use rates. METHODS: A retrospective cohort study was conducted, including 15,015 NICU admissions from an integrated delivery network, across six hospitals over eight years (50% Level III and 50% Level II) computing antibiotic use rates (AURs) stratified by usage: in the first few days of the stay vs. later in the stay and by gestational age. Several metrics were examined for assumptions of strong correlation with AUR: (1) the percentage of infants given antibiotics early in their stays and (2) durations of courses of antibiotics. RESULTS: Results conclude a wide variation in AURs and trends that these rates followed over time. However, there was a decrease in overall AUR from 15.7-16.6 to 10.1-10.8%, with four of the six NICUs recording statistically significant reductions in AUR vs. their first year of measurement. Specifically, the level III NICUs overall AUR decreases from 15.1-16.22 to 8.6-9.4%, and level II NICUs overall AUR 20.3-24.4 to 14.1-16.1%. A particularly successful level II NICU decreased its AUR from 22.9-30.6 to 5.9-9.4%. CONCLUSION: To our knowledge, this is the first study to utilize data analytics at an IDN level to identify trends in AUR, We have identified practices that allowed an institution to reduce NICU AURs significantly, and which, if done as a standard practice, could be replicated on a broader scale.

A Retrospective Cohort Study of Clinical Factors Associated with Transitions of Care among COVID-19 Patients.

Coronavirus Disease 2019 (COVID-19) is an international health crisis. In this article, we report on patient characteristics associated with care transitions of: 1) hospital admission from the emergency department (ED) and 2) escalation to the intensive care unit (ICU). Analysis of data from the electronic medical record (EMR) was performed for patients with COVID-19 seen in the ED of a large Western U.S. Health System from April to August of 2020, totaling 10,079 encounters. Of these, 5172 resulted in admission as an inpatient within 72 h. Inpatient encounters (n = 6079) were also considered for patients with positive COVID-19 test results, of which 970 resulted in a transfer to the ICU or in-hospital mortality. Laboratory results, vital signs, symptoms, and comorbidities were investigated for each of these care transitions. Different top risk factors were found, but two factors common to hospital admission and ICU transfer were respiratory rate and the need for oxygen support. Comorbidities common to both settings were cerebrovascular disease and congestive heart failure. Regarding laboratory results, the neutrophil-to-lymphocyte ratio was associated with transitions to higher levels of care, along with the ratio of aspartate aminotransferase (AST) to alanine aminotransferase (ALT).

Evolution of Multivalent Nanoparticle Adhesion via Specific Molecular Interactions.

The targeted delivery of nanoparticle carriers holds tremendous potential to transform the detection and treatment of diseases. A major attribute of nanoparticles is the ability to form multiple bonds with target cells, which greatly improves the adhesion strength. However, the multivalent binding of nanoparticles is still poorly understood, particularly from a dynamic perspective. In previous experimental work, we studied the kinetics of nanoparticle adhesion and found that the rate of detachment decreased over time. Here, we have applied the adhesive dynamics simulation framework to investigate binding dynamics between an antibody-conjugated, 200-nm-diameter sphere and an ICAM-1-coated surface on the scale of individual bonds. We found that nano adhesive dynamics (NAD) simulations could replicate the time-varying nanoparticle detachment behavior that we observed in experiments. As expected, this behavior correlated with a steady increase in mean bond number with time, but this was attributed to bond accumulation only during the first second that nanoparticles were bound. Longer-term increases in bond number instead were manifested from nanoparticle detachment serving as a selection mechanism to eliminate nanoparticles that had randomly been confined to lower bond valencies. Thus, time-dependent nanoparticle detachment reflects an evolution of the remaining nanoparticle population toward higher overall bond valency. We also found that NAD simulations precisely matched experiments whenever mechanical force loads on bonds were high enough to directly induce rupture. These mechanical forces were in excess of 300 pN and primarily arose from the Brownian motion of the nanoparticle, but we also identified a valency-dependent contribution from bonds pulling on each other. In summary, we have achieved excellent kinetic consistency between NAD simulations and experiments, which has revealed new insights into the dynamics and biophysics of multivalent nanoparticle adhesion. In future work, we will leverage the simulation as a design tool for optimizing targeted nanoparticle agents.

Novel insights from 3D models: the pivotal role of physical symmetry in epithelial organization.

3D tissue culture models are utilized to study breast cancer and other pathologies because they better capture the complexity of in vivo tissue architecture compared to 2D models. However, to mimic the in vivo environment, the mechanics and geometry of the ECM must also be considered. Here, we studied the mechanical environment created in two 3D models, the overlay protocol (OP) and embedded protocol (EP). Mammary epithelial acini features were compared using OP or EP under conditions known to alter acinus organization, i.e. collagen crosslinking and/or ErbB2 receptor activation. Finite element analysis and active microrheology demonstrated that OP creates a physically asymmetric environment with non-uniform mechanical stresses in radial and circumferential directions. Further contrasting with EP, acini in OP displayed cooperation between ErbB2 signalling and matrix crosslinking. These differences in acini phenotype observed between OP and EP highlight the functional impact of physical symmetry in 3D tissue culture models.