Publications

In the dynamic hospital setting, decision support can be a valuable tool for improving patient outcomes. Data-driven inference of future outcomes is challenging in this dynamic setting, where long sequences such as laboratory tests and medications are updated frequently. This is due in part to heterogeneity of data types and mixed-sequence types contained in variable length sequences. In this work we design a probabilistic unsupervised model for multiple arbitrary-length sequences contained in hospitalization Electronic Health Record (EHR) data. The model uses a latent variable structure and captures complex relationships between medications, diagnoses, laboratory tests, neurological assessments, and medications. It can be trained on original data, without requiring any lossy transformations or time binning. Inference algorithms are derived that use partial data to infer properties of the complete sequences, including their length and presence of specific values. We train this model on data from subjects receiving medical care in the Kaiser Permanente Northern California integrated healthcare delivery system. The results are evaluated against held-out data for predicting the length of sequences and presence of Intensive Care Unit (ICU) in hospitalization bed sequences. Our method outperforms a baseline approach, showing that in these experiments the trained model captures information in the sequences that is informative of their future values.

Transfusion was not associated with a decrease in mean cardiac output or mean heart rate. The increase in mean oxygen delivery following transfusion was associated with an increase in mean oxygen consumption after transfusion, especially in patients with sepsis.

Use of a continuous glucose monitoring-based strategy decreased the incidence and severity of hypoglycemia, thus improving the safety of glycemic control.

Multiple experimental and human trials have shown that microcirculatory alterations are frequent in sepsis. In this review, we discuss the characteristics of these alterations, the various mechanisms potentially involved, and the implications for therapy. Sepsis-induced microvascular alterations are characterized by a decrease in capillary density with an increased number of stopped-flow and intermittent-flow capillaries, in close vicinity to well-perfused capillaries. Accordingly, the surface available for exchange is decreased but also is highly heterogeneous. Multiple mechanisms may contribute to these alterations, including endothelial dysfunction, impaired inter-cell communication, altered glycocalyx, adhesion and rolling of white blood cells and platelets, and altered red blood cell deformability. Given the heterogeneous nature of these alterations and the mechanisms potentially involved, classical hemodynamic interventions, such as fluids, red blood cell transfusions, vasopressors, and inotropic agents, have only a limited impact, and the microcirculatory changes often persist after resuscitation. Nevertheless, fluids seem to improve the microcirculation in the early phase of sepsis and dobutamine also can improve the microcirculation, although the magnitude of this effect varies considerably among patients. Finally, maintaining a sufficient perfusion pressure seems to positively influence the microcirculation; however, which mean arterial pressure levels should be targeted remains controversial. Some trials using vasodilating agents, especially nitroglycerin, showed promising initial results but they were challenged in other trials, so it is difficult to recommend the use of these agents in current practice. Other agents can markedly improve the microcirculation, including activated protein C and antithrombin, vitamin C, or steroids. In conclusion, microcirculatory alterations may play an important role in the development of sepsis-related organ dysfunction. At this stage, therapies to target microcirculation specifically are still being investigated.