No abstract

In this paper we introduce RL-CD, a method for solving reinforcement learning problems in non-stationary environments. The method is based on a mechanism for creating, updating and selecting one among several partial models of the environment. The partial models are incrementally built according to the system's capability of making predictions regarding a given sequence of observations. We propose, formalize and show the efficiency of this method both in a simple non-stationary environment and in a noisy scenario. We show that RL-CD performs better than two standard reinforcement learning algorithms and that it has advantages over methods specifically designed to cope with non-stationarity. Finally, we present known limitations of the method and future works.

Background Coagulation abnormalities in COVID-19 patients have not been addressed in depth. Objective To perform a longitudinal evaluation of coagulation profile of patients admitted to the ICU with COVID-19. Methods Conventional coagulation tests, rotational thromboelastometry (ROTEM), platelet function, fibrinolysis, antithrombin, protein C and S were measured at days 0, 1, 3, 7 and 14. Based on median total maximum SOFA score, patients were divided in two groups: SOFA ≤ 10 and SOFA > 10. Results Thirty patients were studied. Some conventional coagulation tests, as aPTT, PT and INR remained unchanged during the study period, while alterations on others coagulation laboratory tests were detected. Fibrinogen levels were increased in both groups. ROTEM maximum clot firmness increased in both groups from Day 0 to Day 14. Moreover, ROTEM–FIBTEM maximum clot firmness was high in both groups, with a slight decrease from day 0 to day 14 in group SOFA ≤ 10 and a slight increase during the same period in group SOFA > 10. Fibrinolysis was low and decreased over time in all groups, with the most pronounced decrease observed in INTEM maximum lysis in group SOFA > 10. Also, D-dimer plasma levels were higher than normal reference range in both groups and free protein S plasma levels were low in both groups at baseline and increased over time, Finally, patients in group SOFA > 10 had lower plasminogen levels and Protein C ​​than patients with SOFA <10, which may represent less fibrinolysis activity during a state of hypercoagulability. Conclusion COVID-19 patients have a pronounced hypercoagulability state, characterized by impaired endogenous anticoagulation and decreased fibrinolysis. The magnitude of coagulation abnormalities seems to correlate with the severity of organ dysfunction. The hypercoagulability state of COVID-19 patients was not only detected by ROTEM but it much more complex, where changes were observed on the fibrinolytic and endogenous anticoagulation system.

7 formalin-inactivated bacteria·mL -1 , for 20 minutes. Vaccinated fish groups presented higher hematocrit, number of erythrocytes and leukocytes than the non-vaccinated group. Serum agglutination titer of intraperitoneally vaccinated fish was higher on both evaluation periods for the three bacteria strains. Only on day 21 post-vaccination fish from the oral and immersion vaccination groups presented higher serum agglutination titer than the non-vaccinated fish for A. hidrophyla and E. durans. Serum antimicrobial activity in vaccinated fish was higher for P. aeroginosa and E. coli than in non-vaccinated fish on both evaluation periods. The different vaccine administration routes stimulated hematological and immunological responses in Nile tilapia 21 days post-vaccination, but intraperitoneal vaccination presented higher total number of leukocytes, lymphocytes and serum agglutination titer.

<b><i>Background:</i></b> Critically ill patients with COVID-19 may develop multiple organ dysfunction syndrome, including acute kidney injury (AKI). We report the incidence, risk factors, associations, and outcomes of AKI and renal replacement therapy (RRT) in critically ill COVID-19 patients. <b><i>Methods:</i></b> We performed a retrospective cohort study of adult patients with COVID-19 diagnosis admitted to the intensive care unit (ICU) between March 2020 and May 2020. Multivariable logistic regression analysis was applied to identify risk factors for the development of AKI and use of RRT. The primary outcome was 60-day mortality after ICU admission. <b><i>Results:</i></b> 101 (50.2%) patients developed AKI (72% on the first day of invasive mechanical ventilation [IMV]), and thirty-four (17%) required RRT. Risk factors for AKI included higher baseline Cr (OR 2.50 [1.33–4.69], <i>p</i> = 0.005), diuretic use (OR 4.14 [1.27–13.49], <i>p</i> = 0.019), and IMV (OR 7.60 [1.37–42.05], <i>p</i> = 0.020). A higher C-reactive protein level was an additional risk factor for RRT (OR 2.12 [1.16–4.33], <i>p</i> = 0.023). Overall 60-day mortality was 14.4% {23.8% (<i>n</i> = 24) in the AKI group versus 5% (<i>n</i> = 5) in the non-AKI group (HR 2.79 [1.04–7.49], <i>p</i> = 0.040); and 35.3% (<i>n</i> = 12) in the RRT group versus 10.2% (<i>n</i> = 17) in the non-RRT group, respectively (HR 2.21 [1.01–4.85], <i>p</i> = 0.047)}. <b><i>Conclusions:</i></b> AKI was common among critically ill COVID-19 patients and occurred early in association with IMV. One in 6 AKI patients received RRT and 1 in 3 patients treated with RRT died in hospital. These findings provide important prognostic information for clinicians caring for these patients.