Coupling the K-nearest neighbors and locally weighted linear regression with ensemble Kalman filter for data-driven data assimilation

Machine learning-based data-driven methods are increasingly being used to extract structures and essences from the ever-increasing pool of geoscience-related big data, which are often used in relation to the atmosphere, oceans, and land surfaces. This study focuses on applying a data-driven forecast model to the classical ensemble Kalman filter process to reconstruct, analyze, and elucidate the model. In this study, a nonparametric sampler from a catalog of historical datasets, namely, a nearest neighbor or analog sampler, is given by numerical simulations. Based on this catalog (sampler), the dynamics physics model is reconstructed using the K-nearest neighbors algorithm. The optimal values of the surrogate model are found, and the forecast step is performed using locally weighted linear regression. Several numerical experiments carried out using the Lorenz-63 and Lorenz-96 models demonstrate that the proposed approach performs as good as the ensemble Kalman filter for larger catalog sizes. This approach is restricted to the ensemble Kalman filter form. However, the basic strategy is not restricted to any particular version of the Kalman filter. It is found that this combined approach can outperform the generally used sequential data assimilation approach when the size of the catalog is substantially large.

An Advanced Statistical Approach Using Weighted Linear Regression in Electroanalytical Method Development for Epinephrine, Uric Acid and Ascorbic Acid Determination

In this study, the use of weighted linear regression in the development of electrochemical methods for the determination of epinephrine (EP), ascorbic acid (AA), and uric acid (UA) is presented. The measurements were performed using a glassy carbon electrode and square-wave voltammetry (SWV). All electroanalytical methods were validated by determination of the limit of detection, limit of quantification, linear concentration range, accuracy, and precision. The normal distribution of all data sets was checked using the quantile-quantile plot and Kolmogorov-Smirnov statistical tests. The heteroscedasticity of the data was tested using Hartley’s test, Bartlett’s test, Cochran’s C test, and the analysis of residuals. The heteroscedastic behavior was observed with all analytes, justifying the use of weighted linear regression. Six different weighting factors were tested, and the best weighted model was determined using relative percentage error. Such statistical approach improved the regression models by giving greater weight on the values with the smallest error and vice versa. Consequently, accuracy of the analytical results (especially in the lower concentration range) was improved. All methods were successfully used for the determination of these analytes in real samples: EP in an epinephrine auto-injector, AA in a dietary supplement, and UA in human urine. The accuracy and precision of real sample analysis using best weighted model gave satisfactory results with recoveries between 95.21–113.23% and relative standard deviations between 0.85–7.98%. The SWV measurement takes about 40 s, which makes the presented methods for the determination of EP, AA, and UA a promising alternative to chromatographic techniques in terms of speed, analysis, and equipment costs, as the analysis is performed without organic solvents.

Regression with Gaussian Mixture ModelsApplied to Track Fitting

This note describes the application of Gaussian mixture regression to track fitting with a Gaussian mixture model of the position errors. The mixture model is assumed to have two components with identical component means. Under the premise that the association of each measurement to a specific mixture component is known, the Gaussian mixture regression is shown to have consistently better resolution than weighted linear regression with equivalent homoskedastic errors. The improvement that can be achieved is systematically investigated over a wide range of mixture distributions. The results confirm that with constant homoskedastic variance the gain is larger for larger mixture weight of the narrow component and for smaller ratio of the width of the narrow component and the width of the wide component.

ANTIBIOTIC USE IN SLOVENIAN HOSPITALS

Motivation/Background: Antibiotics are commonly overused and misused what increase the emergence of resistant organisms, side- effects and costs. To assess the appropriate use of antibiotics many methods are available. The aim of the present study is to find correlation between antibiotic use and case mix index (CMI) in Slovenian hospitals. Method: In retrospective study (in the years between 2004 and 2013) we correlated the total consumption of antibiotics for systemic use and CMI. Weighted linear regression test analysis was performed to determine correlation between defined daily dose (DDD) / 100 admissions and DDD / 100 bed-days and CMI. Results: The total antibiotic consumption in all included hospitals was in mean 317.69 DDD / 100 admissions and 58.88 DDD / 100 bed days, respectively. CMI range were from 1.25 to 3.55. A significant correlation between consumption expressed in DDD / 100 admissions and CMI (p = 0.028) and DDD / 100 bed days and CMI (p =0.008) was found. Conclusions: Thus, detailed analysis of correlations between DDD of antibiotics and CMI may constitutes a proper use of antibiotics.