Treating missing data in a clinical neuropsychological dataset--data imputation

Missing data frequently reduce the applicability of clinically collected data in research requiring multivariate statistics. In data imputation, missing values are replaced by predicted values obtained from models based on auxiliary information. Our aim was to complete a clinical child neuropsychological data set containing 5.2% of missing observations. This was to be used in research requiring multivariate statistics. We compared four data imputation methods by artificially deleting some data. A real-donor imputation method which preserved the parameter estimates and which predicted the observed values with acceptable accuracy was used to complete the data set. In addressing the lack of studies with regard to treatment of missing data in neuropsychological data sets, this study presents information on the outcomes of applying data imputation methods to such data. The imputation modeling described can be applied to a variety of clinical neuropsychological data sets.

Serum homocysteine at 16 weeks and subsequent preeclampsia

OBJECTIVE

To determine whether elevated homocysteine levels precede the development of preeclampsia.

METHODS

Study subjects were selected from a population-based cohort of 1049 nulliparous women from whom serum was collected for Down syndrome screening at 16 weeks' gestation. For 34 women who developed preeclampsia, 68 control women were chosen who remained normotensive. Homocysteine was analyzed by high-performance liquid chromatography and fluorescence detection. The sample size allowed detection of a 1.25-micromol/L difference at a significance level of 0.05 and the power of 0.81.

RESULTS

At 16 weeks' gestation, concentrations (mean, 95% confidence interval) of homocysteine in women who developed preeclampsia, 6.99 (6.42, 7.55) micromol/L, were similar to those who remained normotensive, 6.91 (6.45, 7.34) micromol/L.

CONCLUSION

Significant changes in homocysteine metabolism did not predate the appearance of clinical preeclampsia.

Serum insulin-like growth factor binding protein-1 at 16 weeks and subsequent preeclampsia

OBJECTIVE

To determine whether serum concentrations of insulin-like growth factor-binding protein-1 (IGFBP-1), a major decidual protein, at 16 weeks' gestation differ between women who later develop pregnancy-related hypertension and normotensive women.

METHODS

Concentrations of IGFBP-1 were measured using immunoenzymometric assay in serum samples collected for alpha-fetoprotein (AFP) and free beta subunit of hCG (free beta-hCG) determinations in a Down syndrome screening program at 16 weeks' gestation in a population-based cohort of 1049 nulliparous women. After exclusion of subjects with multiple pregnancies, insulin-dependent diabetes, major fetal malformations, and incomplete data, 917 subjects remained eligible.

RESULTS

The mean levels (+/- standard deviation) of IGFBP-1 were significantly lower in 34 women who later developed preeclampsia (73 +/- 43 microg/L, P < .01) and in 80 women with White A diabetes (84.7 +/- 53 microg/L, P < .01) compared with controls (103 +/- 58 microg/L). In seven women with White A diabetes and subsequent preeclampsia IGFBP-1 levels were especially low (41 +/- 34 microg/L). The concentrations of AFP and free beta-hCG in the subgroups with hypertensive disorders were not significantly different from those of normotensive women.

CONCLUSION

Decreased IGFBP-1 levels at 16 weeks' gestation in women who develop preeclampsia might indicate impaired decidual function. Hyperinsulinemia, a known risk factor for preeclampsia, might contribute to decreased concentrations of serum IGFBP-1. However, due to low sensitivity, assay of serum IGFBP-1 was not clinically valuable for predicting preeclampsia.

Volume rendering using seed-filling acceleration: supporting cut planes by fast reseeding

OBJECTIVE

Seed filling in view lattice is a method for accelerating volume rendering. Our previously published results on using seed filling had the limitation that they did not support cut planes. Cut planes are one of the main advantages of using volume rendering over surface rendering. In this article we describe the seed-filling acceleration technique and propose two fast reseeding processes that may be performed for each frame.

MATERIALS AND METHODS

The reseeding process performs a connected component search on the cut plane within the volume data set. The amount of processing done by a connected component search algorithm is proportional to N(2), assuming the data set size is N(3). We implemented two connected component search algorithms: one for orthogonal cut planes and another for oblique cut planes. We measured the time taken by the cutting and estimated the effect on the rendering performance when the cut plane was moved interactively.

RESULTS

Our measurements show that rendering rates of several frames per second can be achieved with a 266-MHz Pentium II PC, even when the object is interactively modified with cut planes during the rendering.

CONCLUSIONS

We have shown that the seed-filling algorithm is also applicable in situations where the displayed data is modified interactively using cut planes or objects. Applications in this regard include, for example, computer-aided surgery systems, where the instrument may be used to interactively perform the simulated cut in the data set.

Effect of calcitonin on the alcohol drinking of rats

Previous work has shown that calcitonin inhibits eating by rats and that it affects several neurotransmitter systems suspected to play a role in alcohol consumption. The present study was an initial test of whether calcitonin does affect voluntary alcohol consumption by male Wistar rats with prolonged alcohol experience. Calcitonin (20 IU/kg) or saline was injected subcutaneously on 10 consecutive days when the rats (n = 20) had continual access to 10% (v/v) ethanol solution, and to food and water. Using a cross-over design, the effects of 40 IU/kg calcitonin vs. saline were then examined in a second 10-day treatment period. Similar patterns of effects were obtained with both calcitonin doses, but the patterns differed with alcohol, food, and water intake. Alcohol drinking showed biphasic changes with both doses, producing highly significant Treatment x Day interactions (p < 1E-10 and p = 6E-7): it was significantly reduced on the first day of calcitonin treatment and significantly increased on the last few days. Food intake was reduced on all calcitonin days although most markedly on the first. Water drinking was not altered on the first calcitonin day, but was greatly increased on the second, then gradually returned toward the baseline. In a second experiment, the animals were switched to 1 hr of alcohol access per day, and calcitonin (20 IU/kg) was administered periodically to one group 4 hr before the alcohol access. Alcohol drinking was significantly reduced in all cases when the calcitonin injection was preceded by at least 1 day without calcitonin.(ABSTRACT TRUNCATED AT 250 WORDS)