R. Choudary Hanumara

Improving forecasting for telemarketing centers by ARIMA modeling with intervention

Abstract In this study we analyze existing and improved methods for forecasting incoming calls to telemarketing centers for the purposes of planning and budgeting. We analyze the use of additive and multiplicative versions of Holt–Winters (HW) exponentially weighted moving average models and compare it to Box–Jenkins (ARIMA) modeling with intervention analysis. We determine the forecasting accuracy of HW and ARIMA models for samples of telemarketing data. Although there is much evidence in recent literature that “simple models” such as Holt–Winters perform as well as or better than more complex models, we find that ARIMA models with intervention analysis perform better for the time series studied.

White Matter Injury after Cerebral Ischemia in Ovine Fetuses

The effects of cerebral ischemia on white matter changes in ovine fetuses were examined after exposure to bilateral carotid artery occlusion. Fetal sheep were exposed to 30 min of ischemia followed by 48 (I/R-48, n = 8) or 72 (I/R-72, n = 10) h of reperfusion or control sham treatment (control, n = 4). Serial coronal sections stained with Luxol fast blue/hematoxylin and eosin were scored for white matter, cerebral cortical, and hippocampal lesions. All areas received graded pathologic scores of 0 to 5, reflecting the degree of injury where 0 = 0%, 1 = 1% to 25%, 2 = 26% to 50%, 3 = 51% to 75%, 4 = 76% to 95%, and 5 = 96% to 100% of the area damaged. Dual-label immunofluorescence using antibodies against glial fibrillary acidic protein (GFAP) and myelin basic protein (MBP) were used to characterize white matter lesions. Basic fibroblast growth factor (FGF-2) was measured in the frontal cortex by ELISA. Results of the pathologic scores showed that the white matter of the I/R-72 (2.74 ± 0.53, mean ± SEM) was more (p < 0.05) damaged when compared with the control (0.80 ± 0.33) group. Cortical lesions were greater (p < 0.05) in the I/R-48 (2.12 ± 0.35) than the control (0.93 ± 0.09) group. White matter lesions were characterized by reactive GFAP-positive astrocytes and a loss of MBP in oligodendrocytes. The ratio of MBP to GFAP decreased (p < 0.05) as a function of ischemia, indicative of a proportionally greater loss of MBP than GFAP. FGF-2 concentrations were higher (p < 0.05) in the I/R-72 than the control group and there was a direct correlation between the pathologic scores (PS) and FGF-2 concentrations (FGF-2 = e(1.6 PS-0.90) + 743, n = 17, r = 0.73, p < 0.001). We conclude that carotid artery occlusion results in quantifiable white matter lesions that are associated with a loss of MBP from myelin, and that FGF-2, a purported mediator of recovery from brain injury in adult subjects, increases in concentration in proportion to the severity of brain damage in the fetus.

An empirical comparison of a fit of linear and non-linear models for seasonal growth in fish

Abstract Modifications to the von Bertalanffy model of body growth by incorporating time-varying coefficients in the form of trigonometric functions have been proposed by several authors. In particular, Pauly and Gaschutz presented a non-linear model and transformed it to the form of a multiple linear regression model for fitting by standard methods. Approximate expressions for the variances and covariances of the estimates of the parameters thus obtained have been derived in this paper. Non-linear methods of fitting seasonal growth models are only moderately difficult in computational aspects, do not require prior estimates of any parameters, and offer flexibility in choosing natural error structures. The means and the covariance matrices of the estimates of the parameters obtained from non-linear and linear methods are compared. Simulation studies indicate that the estimates obtained are distributed approximately multivariate normal, thus facilitating the comparison of growth curves.

Effects of systematic error, estimates and uncertainties in chemical mass balance apportionments: Quail Roost II revisited

Abstract The Quail Roost II synthetic data set II was used to derive a comprehensive method of estimating uncertainties for chemical mass balance (CMB) apportionments. Collinearity-diagnostic procedures were applied to CMB apportionments of data set II to identify seriously collinear source profiles and evaluate the effects of the degree of collinearity on source-strength estimates and their uncertainties. Fractional uncertainties of CMB estimates were up to three times higher for collinear source profiles than for independent ones. A theoretical analysis of CMB results for synthetic data set II led to the following general conclusions about CMB methodology. Uncertainties for average estimated source strengths will be unrealistically low unless sources whose estimates are constrained to zero are included when calculating uncertainties. Covariance in source-strength estimates is caused by collinearity and systematic errors in source specification and composition. Propagated uncertainties may be underestimated unless covariances as well as variances of estimates are included. Apportioning the average aerosol will account for systematic errors only when the correct model is known, when measurement uncertainties in ambient and source-profile data are realistic, and when the source profiles are not collinear.

Dose Spacing in Early Dose Response Clinical Trial Designs

Finding the right dose or the right range of doses is one of the most important objectives in the clinical development program of a new drug. In designing early Phase 2 dose response studies, one critical question is: “What spacing should be used between test doses?” In this article, we propose an intuitive, model-free approach to address this question. Simulations are performed to compare the proposed design with optimal designs under logistic and normal models. The proposed method for dose allocation can be applied to various types of dose response studies.

Effects of exogenous glucose on brain ischemia in ovine fetuses

We examined the effects of prolonged moderate hyperglycemia with and without an additional rapid glucose injection on ischemic brain injury in the fetus. Twenty-five ewes (117–124 d of gestation) were assigned to one of four groups: 1) glucose-infused fetuses exposed to 30 min of carotid artery occlusion followed by 48 h of reperfusion (I/R-Glu, n = 8); 2) glucose-infused plus rapid glucose injection given 100 min before 30 min of occlusion followed by 48 h of reperfusion (I/R-GluR, n = 4); 3) placebo-infused exposed to 30 min of occlusion and 48 h of reperfusion (I/R-PL, n = 8); and 4) glucose-infused sham occlusion and 48 h of sham reperfusion (control, n = 5). After baseline measurements, fetuses were infused with glucose (9–16 mg/kg/min) for 48 h before and after carotid occlusion or sham treatment. The I/R-PL group received 0.9% NaCl. Brain pathologic outcome was determined. Serial sections stained with Luxol fast blue-hematoxylin and eosin were scored for white matter, cerebral cortical, and hippocampal lesions. These areas received graded pathologic scores of 0 to 5, reflecting the amount of injury, where 0 = 0%, 1 = 1–25%, 2 = 26–50%, 3 = 51–75%, 4 = 76–95%, and 5 = 96–100% of the area damaged. Comparisons of the pathologic scores for cerebral cortex (CC), white matter (WM), and hippocampus (H) demonstrated that the I/R-GluR (CC: 4.56 ± 0.11, WM: 4.50 ± 0.11, H: 3.44 ± 0.48, mean ± SEM) had more (p < 0.05) damage than the I/R-Glu (CC: 2.46 ± 0.47, WM: 1.97 ± 0.37, H: 1.81 ± 0.36) and control (CC: 1.12 ± 0.13, WM: 0.82 ± 0.34, H: 0.80 ± 0.34) groups. The pathologic scores in the I/R-Glu were (p < 0.05) greater than the control, but not the I/R-PL (CC: 2.12 ± 0.35, WM: 2.20 ± 0.44, H: 1.59 ± 0.41) group. We conclude that exposure to prolonged moderate hyperglycemia before ischemia and during reperfusion does not affect the extent of brain injury, but exposure to an additional acute increase in plasma glucose concentration before ischemia is extremely detrimental to the fetal brain.

Comparative effects of glucose- and mannitol-induced osmolar stress on blood-brain barrier function in ovine fetuses and lambs

We examined the effects of hyperglycemic hyperosmolality on blood-brain barrier (BBB) permeability during development. We hypothesized that the barrier becomes more resistant to hyperglycemic hyperosmolality during development, and the immature BBB is more resistant to glucose than to mannitol hyperosmolality. We quantified the BBB response to hyperosmolality with the blood-to-brain transfer constant (Ki) in immature fetuses, premature, and newborn lambs. Ki increased as a function of increases in osmolality. A segmented regression model described the relationship between Ki and osmolality. At lower osmolalities, changes in Ki were minimal but after a threshold, increases were linear. We examined responses of Ki to hyperglycemic hyperosmolality by comparing the thresholds and slopes of the second regression segments. Lower thresholds and steeper slopes indicate greater vulnerability to hyperosmolality. Thresholds increased (P < 0.05) during development in pons and superior colliculus. Thresholds were higher (P < 0.05) during glucose than mannitol hyperosmolality in thalamus, superior colliculus, inferior colliculus and medulla of premature lambs, and in cerebrum and cerebellum of newborns. We conclude that BBB permeability increased as a function of changes in glucose osmolality, the barrier becomes more resistant to glucose hyperosmolality in two brain regions during development, and the barrier is more resistant to glucose than to mannitol hyperosmolality in some brain regions of premature and newborn lambs.

A comparison of non-linear growth models for fisheries

Abstract Several extensions to the classical von Bertalanffy model, based on the consideration that the growth-time relationships in fishes in temperate and polar areas vary seasonally, have appeared in the literature. In this paper, we compare the seasonal growth models of Cloern and Nichols, Hoenig and Hanumara, Pauly and Gaschutz, and Pitcher and MacDonald using several data sets. For each model/data set combination, measures of intrinsic non-linearity and parameter effects non-linearity as proposed by Bates and Watts are computed. The model parameter estimates are evaluated using bias, skewness and kurtosis to determine which of them contribute to the non-linear behaviour of the model. These results show that the Pauly and Gaschutz and Hoenig and Hanumara models exhibit properties closely resembling that of a linear model, and hence are selected over the other models.

Estimating Imprecisions of Measuring Instruments

In some experiments, simultaneous measurements on an item by several instruments may be obtained. Procedures for simultaneous confidence intervals of the vltriance in errors of measurement or the imprecisions of the instruments are disclwcd. When the number of instruments is three, non-negative estimates of variances are proposed.

A table of some percentage points of the distribution of a difference between independent chi-square variables

The distribution of Bell-Doksum measure of correlation is that of a difference between independent chi-square variables with equal weights. A table of percentage points computed here for the distribution may be used to test a hypothesis of no correlation between two variables. The distribution of a diffference between independent chi-square variables is also useful in studying variance component estimators and some general results corresponding to the distribution are given.