A kernel-based two-class classifier for imbalanced data sets

Many kernel classifier construction algorithms adopt classification accuracy as performance metrics in model evaluation. Moreover, equal weighting is often applied to each data sample in parameter estimation. These modeling practices often become problematic if the data sets are imbalanced. We present a kernel classifier construction algorithm using orthogonal forward selection (OFS) in order to optimize the model generalization for imbalanced two-class data sets. This kernel classifier identification algorithm is based on a new regularized orthogonal weighted least squares (ROWLS) estimator and the model selection criterion of maximal leave-one-out area under curve (LOO-AUC) of the receiver operating characteristics (ROCs). It is shown that, owing to the orthogonalization procedure, the LOO-AUC can be calculated via an analytic formula based on the new regularized orthogonal weighted least squares parameter estimator, without actually splitting the estimation data set. The proposed algorithm can achieve minimal computational expense via a set of forward recursive updating formula in searching model terms with maximal incremental LOO-AUC value. Numerical examples are used to demonstrate the efficacy of the algorithm.

Construction of neurofuzzy models for imbalanced data classification

We propose a new class of neurofuzzy construction algorithms with the aim of maximizing generalization capability specifically for imbalanced data classification problems based on leave-one-out (LOO) cross validation. The algorithms are in two stages, first an initial rule base is constructed based on estimating the Gaussian mixture model with analysis of variance decomposition from input data; the second stage carries out the joint weighted least squares parameter estimation and rule selection using orthogonal forward subspace selection (OFSS)procedure. We show how different LOO based rule selection criteria can be incorporated with OFSS, and advocate either maximizing the leave-one-out area under curve of the receiver operating characteristics, or maximizing the leave-one-out Fmeasure if the data sets exhibit imbalanced class distribution. Extensive comparative simulations illustrate the effectiveness of the proposed algorithms.

Effect of dietary supplementation of different oils during the first or second half of pregnancy on the glucose tolerance of the sow

Poor glucose tolerance may be an under-researched contributory factor in the high (10% to 20%) pre-weaning mortality rate observed in pigs. Insulin resistance commences at around week 12 of gestation in the sow, although there are conflicting reports in the literature about the extent to which insulin resistance is modulated by maternal diet. The aim of the study was to determine the effects of supplementing the maternal diet with different dietary oils during either the first half or the second half of gestation on the glucose tolerance of the sow. Sows were offered the control (C: n = 5) diet as pellets or the C diet plus 10% extra energy (h = 16 per group) derived from either. (i) extra pellets; (ii) palm oil; (iii) olive oil; (iv) sunflower oil; or (v) fish oil. Experimental diets were fed during either the first (G1) or second (G2) half of gestation. A glucose tolerance test (GTT) was conducted on day 108 of gestation by administering 0.5g/kg glucose i.v. Blood samples were taken every 5 to 10 min for 90 min post administration. The change in body weight and backfat thickness during gestation was similar but both type and timing of dietary supplementation influenced litter size and weight. With the exception of the sunflower oil group, supplementing the maternal diet in G1 resulted in larger and heavier litters, particularly in mothers offered palm oil. Basal blood glucose concentrations tended to be more elevated in G1 than G2 groups, whilst plasma insulin concentrations were similar Following a GTT, the adjusted area under the curve was greater in G1 compared to G2 sows, despite no differences in glucose clearance. Maternal diet appeared to influence the relationship between glucose curve characteristics following a GTT and litter outcome. In conclusion, the degree of insulin sensitivity can be altered by both the period during which maternal nutritional supplementation is offered and the fatty acid profile of the diet.

Comparison of sample size formulae for 2x2 cross-over designs applied to bioequivalence studies

We consider the comparison of two formulations in terms of average bioequivalence using the 2 × 2 cross-over design. In a bioequivalence study, the primary outcome is a pharmacokinetic measure, such as the area under the plasma concentration by time curve, which is usually assumed to have a lognormal distribution. The criterion typically used for claiming bioequivalence is that the 90% confidence interval for the ratio of the means should lie within the interval (0.80, 1.25), or equivalently the 90% confidence interval for the differences in the means on the natural log scale should be within the interval (-0.2231, 0.2231). We compare the gold standard method for calculation of the sample size based on the non-central t distribution with those based on the central t and normal distributions. In practice, the differences between the various approaches are likely to be small. Further approximations to the power function are sometimes used to simplify the calculations. These approximations should be used with caution, because the sample size required for a desirable level of power might be under- or overestimated compared to the gold standard method. However, in some situations the approximate methods produce very similar sample sizes to the gold standard method. Copyright © 2005 John Wiley & Sons, Ltd.

Ingestion of onion soup high in quercetin inhibits platelet aggregation and essential components of the collagen-stimulated platelet activation pathway in man: a pilot study

Epidemiological data suggest that those who consume a diet rich in quercetin-containing foods may have a reduced risk of CVD. Furthermore, in vitro and ex vivo studies have observed the inhibition of collagen-induced platelet activation by quercetin. The aim of the present study was to investigate the possible inhibitory effects of quercetin ingestion from a dietary source on collagen-stimulated platelet aggregation and signalling. A double-blind randomised cross-over pilot study was undertaken. Subjects ingested a soup containing either a high or a low amount of quercetin. Plasma quercetin concentrations and platelet aggregation and signalling were assessed after soup ingestion. The high-quercetin soup contained 69 mg total quercetin compared with the low-quercetin soup containing 5 mg total quercetin. Plasma quercetin concentrations were significantly higher after high-quercetin soup ingestion than after low-quercetin soup ingestion and peaked at 2.59 (SEM 0.42) mu mol/l. Collagen-stimulated (0.5 mu g/ml) platelet aggregation was inhibited after ingestion of the high-quercetin soup in a time-dependent manner. Collagen-stimulated tyrosine phosphorylation of a key component of the collagen-signalling pathway via glycoprotein VI, Syk, was significantly inhibited by ingestion of the high-quercetin soup. The inhibition of Syk tyrosine phosphorylation was correlated with the area under the curve for the high-quercetin plasma profile. In conclusion, the ingestion of quercetin from a dietary source of onion soup could inhibit some aspects of collagen-stimulated platelet aggregation and signalling ex vivo. This further substantiates the epidemiological data suggesting that those who preferentially consume high amounts of quercetin-containing foods have a reduced risk of thrombosis and potential CVD risk.

The effect of potato variety mixtures on epidemics of late blight, in relation to plot size and level of resistance

Potatoes of a number of varieties of contrasting levels of resistance were planted in pure or mixed stands in four experiments over 3 years. Three experiments compared the late blight severity and progress in mixtures with that in pure stands. Disease on susceptible or moderately resistant varieties typical of those in commercial use was similar in mixtures and pure stands. In 2 of 3 years, there were slight reductions on cv. Sante, which is moderately susceptible, in mixture with cv. Cara, which is moderately resistant. Cara was unaffected by this mixture. Mixtures of an immune or near-immune partner with Cara or Sante substantially reduced disease on the latter. The effect of the size of plots of individual varieties or mixtures on blight severity was compared in two experiments. Larger plots had a greater area under the disease progress curve, but the average rate of disease progress was greater in smaller plots; this may be because most disease progress took place later, under more favourable conditions, in the smaller plots. In one experiment, two planting densities were used. Density had no effect on disease and did not interact with mixture effects. The overall conclusion is that, while mixtures of potato varieties may be desirable for other reasons, they do not offer any improvement on the average of the disease resistance of the components.

Long-term monounsaturated fatty acid diets reduce platelet aggregation in healthy young subjects

The aim of the present study was to compare the response of a range of atherogenic and thrombogenic risk markers to two dietary levels of saturated fatty acid (SFA) substitution with monounsaturated fatty acids (MUFA) in students living in a university hall of residence. Although the benefits of such diets have been reported for plasma lipoproteins in high-risk groups, more needs to be known about effects of more modest SFA-MUFA substitutions over the long term and in young healthy adults. In a parallel design over 16 weeks, fifty-one healthy young subjects were randomised to one of two diets: (1) a moderate-MUFA diet in which 16 g dietary SFA/100 g total fatty acids were substituted with MUFA (n 25); (2) a high-MUFA diet in which 33 g dietary SFA/100 g total fatty acids were substituted with MUFA (n 26). All subjects followed an 8-week run-in diet (reference diet), with a fatty acid composition close to the UK average values. There were no differences in plasma lipid responses between the two diets over 16 weeks of the study with similar reductions in total cholesterol (P<0.001) and LDL-cholesterol (P<0.01) in both groups; a small but significant reduction in HDL-cholesterol was also observed in both groups (P<0.01). Platelet responses to ADP (P<0.01) and arachidonic acid (P<0.05) differed with time on the two diets; at 16 weeks, platelet aggregatory response to ADP was significantly lower on the high-MUFA than the moderate-MUFA (P<0.01) diet; ADP responses were also significantly lower within this group at 8 (P< 0.05) and 16 (P< 0.01) weeks compared with baseline. There were no differences in fasting factor VII activity (factors VIII and VIIag), fibrinogen concentration or tissue-type plasminogen activator activity between the diets. There were no differences in postprandial factor VIII responses to a standard meal (area under the curve) between the diets after 16 weeks, but postprandial factor VIII response was lower than on the high-MUFA diet compared with baseline (P<0.01). In conclusion, a high-MUFA diet sustains potentially beneficial effects on platelet aggregation and postprandial activation of factor VII. Moderate or high substitution of MUFA for SFA achieves similar reductions in fasting blood lipids in young healthy subjects.

Chylomicron particle size and number, factor VII activation and dietary monounsaturated fatty acids

This study evaluated the effects of substituting dietary saturated fatty acids (SFAs) with monounsaturated fatty acids (MUFAs) on postprandial chylomicron (triacylglycerol (TAG), apolipoprotein B-48 (apo B-48) and retinyl ester (RE)), chylomicron particle size and factor VII (FVII) response when subjects were given a standard meal. In a controlled sequential design, 51 healthy young subjects followed an SFA-rich diet (Reference diet) for 8 weeks after which half of the subjects followed a moderate MUFA diet (n = 25) and half followed a high MUFA diet (n = 26) for 16 weeks. Fasting lipoprotein and lipid measurements were evaluated at baseline and at 8-week intervals during the Reference and MUFA diets. In 25 of the subjects (n = 12 moderate MUFA, n = 13 high MUFA), postprandial responses to a standard test meal containing RE and 13 C-tripalmitin were investigated at the end of the Reference and the MUFA diet periods. Although there were no differences in the postprandial lipid markers (TAG, RE, C-13-TAG) on the two diets, the postprandial apo B-48 response (incremental area under the curve (IAUC) was reduced by 21% on the moderate MUFA diet (NS) and by 54% on the high MUFA diet (P < 0.01). The postprandial peak concentrations of apo B-48 were reduced by 33% on the moderate MUFA diet (P < 0.01) and 48% on the high MUFA diet (P < 0.001). Fasting values for factor VII activity (FVIIc), activated factor VII (FVIIa) or factor VII antigen (FVIIag) did not differ significantly when subjects were transferred from Reference to MUFA diets. However, the postprandial increases in coagulation FVII activity (FVIIc) were 18% lower and of activated FVII (FVIIa) were 17% lower on the moderate MUFA diet (NS). Postprandial increases in FVIIc and FVIIa were 50% (P < 0.05) and 29% (P < 0.07) lower on the high MUFA diet and the area under the postprandial FVIIc response curve (AUC) was also lower on the high MUFA diet (P < 0.05). Significantly higher ratios of RE:apo B-48 (P < 0.001) and 13 C-palmitic acid:apo B-48 (P < 0.01) during both MUFA diets suggest that the CMs formed carry larger amounts of dietary lipids per particle, reflecting an adaptation to form larger lipid droplets in the enterocyte when increased amounts of dietary MUFAs are fed. Smaller numbers of larger chylomicrons may explain attenuated activation of factor VII during the postprandial state when the background diet is rich in MUFA. (C) 2002 Elsevier Science Ireland Ltd. All rights reserved.

Influence of apoA-V gene variants on postprandial triglyceride metabolism: impact of gender

Although apolipoprotein AN (apoA-V) polymorphisms have been consistently associated with fasting triglyceride (TG) levels, their impact on postprandial lipemia remains relatively unknown. In this study, we investigate the impact of two common apoA-V polymorphisms (-1131 T>C and S19W) and apoA-V haplotypes on fasting and postprandial lipid metabolism in adults in the United Kingdom (n = 259). Compared with the wild-type TT, apoA-V -1131 TC heterozygotes had 15% (P = 0.057) and 21% (P = 0.002) higher fasting TG and postprandial TG area under the curve (AUC), respectively. Significant (P = 0.038) and nearly significant (P = 0.057) gender X genotype interactions were observed for fasting TG and TG AUC, with a greater impact of genotype in males. Lower HDL-cholesterol was associated with the rare TC genotype (P = 0.047). Significant linkage disequilibrium was found between the apoA-V -1131 T>C and the apoC-III 3238 C>G variants, with univariate analysis indicating an impact of this apoC-III single nucleotide polymorphism (SNP) on TG AUC (P = 0.015). However, in linear regression analysis, a significant independent association with TG AUC (P = 0.007) was only evident for the apoA-V -1131 T>C SNP, indicating a greater relative importance of the apoA-V genotype.

Acute effects of meal fatty acids on postprandial NEFA, glucose and apo E response: implications for insulin sensitivity and lipoprotein regulation?

Our aim was to determine whether meal fatty acids influence insulin and glucose responses to mixed meals and whether these effects can be explained by variations in postprandial NEFA and Apo, which regulate the metabolism of triacylglycerol-rich lipoproteins (Apo C and E). A single-blind crossover study examined the effects of single meals enriched in saturated fatty acids SFA), n-6 PUFA and MUFA on plasma metabolite and insulin responses. The triacylglycerol response following the PUFA meal showed a lower net incremental area under the curve than following the SFA and MUFA meals (P < 0.007). Compared with the SFA meal, the PUFA meal showed a lower net incremental area under the curve for the NEFA response from initial suppression to the end of the postprandial period (180-480 min; P < 0.02), and both PUFA and MUFA showed a lower net incremental glucose response (P < 0.02), although insulin concentrations were similar between meals. The pattern of the Apo E response was also different following the SFA meal (P < 0.02). There was a significant association between the net incremental NEFA (180-480 min) and glucose response (r(s)=0.409, P=0.025), and in multiple regression analysis the NEFA response accounted for 24 % of the variation in glucose response. Meal SFA have adverse effects on the postprandial glucose response that may be due to greater elevations in NEFA arising from differences in the metabolism of SFA- v. PUFA- and MUFA-rich lipoproteins. Elevated Apo E responses to high-SFA meals may have important implications for the hepatic metabolism of triacylglycerol-rich lipoproteins.

Physiological mechanisms mediating aspartame-induced satiety

Aspartame has been previously shown to increase satiety. This study aimed to investigate a possible role for the satiety hormones cholecystokinin (CCK) and glucagon-like peptide-1 (GLP-1) in this effect. The effects of the constituents of aspartame, phenylalanine and aspartic acid, were also examined. Six subjects consumed an encapsulated preload consisting of either 400 mg aspartame, 176 mg aspartic acid + 224 mg phenylalanine, or 400 mg corn flour (control), with 1.5 g paracetamol dissolved in 450 ml water to measure gastric emptying. A 1983-kJ liquid meal was consumed 60 min later. Plasma CCK, GLP-1, glucose-dependent insulinotropic polypeptide (GIP), glucose, and insulin were measured over 0-120 min. Gastric emptying was measured from 0 to 60 min. Plasma GLP-1 concentrations decreased following the liquid meal (60-120 min) after both the aspartame and amino acids preloads (control, 2096.9 pmol/l min; aspartame, 536.6 pmol/l min; amino acids, 861.8 pmol/l min; incremental area under the curve [AUC] 60-120 min, P<.05). Desire to cat was reduced from 60 to 120 min following the amino acids preload (control, -337.1 mm min; aspartame, -505.4 mm min; amino acids, -1497.1 mm min; incremental AUC 60-120 min, P<.05). However, gastric emptying rates, plasma CCK, GIP, insulin, and glucose concentrations were unaffected. There was a correlation between the increase in plasma phenylalanine and decrease in desire to eat after the liquid meal following the constituent amino acids (r = -.9774, P=.004). In conclusion, it is unlikely that aspartame increases satiety via CCK- or GLP-1-mediated mechanisms, but small changes in circulating phenylalanine concentrations may influence appetite. (C) 2003 Elsevier Science Inc. All rights reserved.

Casein and whey exert different effects on plasma amino acid profiles, gastrointestinal hormone secretion and appetite

Protein, generally agreed to be the most satiating macronutrient, may differ in its effects on appetite depending on the protein source and variation in digestion and absorption. We investigated the effects of two milk protein types, casein and whey, on food intake and subjective ratings of hunger and fullness, and on postprandial metabolite and gastrointestinal hormone responses. Two studies were undertaken. The first study showed that energy intake from a buffet meal ad libitum was significantly less 90 min after a 1700 kJ liquid preload containing 48 g whey, compared with an equivalent casein preload (P<0.05). In the second study, the same whey preload led to a 28 % increase in postprandial plasma amino acid concentrations over 3 h compared with casein (incremental area under the curve (iAUC), P<0.05). Plasma cholecystokinin (CCK) was increased by 60 % (iAUC, P<0.005), glucagon-like peptide (GLP)-1 by 65 % (iAUC, P<0.05) and glucose-dependent insulinotropic polypeptide by 36 % (iAUC, P<0.01) following the whey preload compared with the casein. Gastric emptying was influenced by protein type as evidenced by differing plasma paracetamol profiles with the two preloads. Greater subjective satiety followed the whey test meal (P<0.05). These results implicate post-absorptive increases in plasma amino acids together with both CCK and GLP-1 as potential mediators of the increased satiety response to whey and emphasise the importance of considering the impact of protein type on the appetite response to a mixed meal.

Efficacy of three screening instruments in the identification of autistic-spectrum disorders

Background Screening instruments for autistic-spectrum disorders have not been compared in the same sample. Aims To compare the Social Communication Questionnaire (SCQ), the Social Responsiveness Scale (SRS) and the Children's Communication Checklist (CCC). Method Screen and diagnostic assessments on 119 children between 9 and 13 years of age with special educational needs with and without autistic-spectrum disorders were weighted to estimate screen characteristics for a realistic target population. Results The SCQ performed best (area under receiver operating characteristic curve (AUC)=0.90; sensitivity. 6; specificity 0.78). The SRS had a lower AUC (0.77) with high sensitivity (0.78) and moderate specificity (0.67). The CCC had a high sensitivity but lower specificity (AUC=0.79; sensitivity 0.93; specificity 0.46). The AUC of the SRS and CCC was lower for children with IQ < 70. Behaviour problems reduced specificity for all three instruments. Conclusions The SCQ, SRS and CCC showed strong to moderate ability to identify autistic-spectrum disorder in this at-risk sample of school-age children with special educational needs.

Characterization and classification of tumor lesions using computerized fractal-based texture analysis and support vector machines in digital mammograms

Objective: This paper presents a detailed study of fractal-based methods for texture characterization of mammographic mass lesions and architectural distortion. The purpose of this study is to explore the use of fractal and lacunarity analysis for the characterization and classification of both tumor lesions and normal breast parenchyma in mammography. Materials and methods: We conducted comparative evaluations of five popular fractal dimension estimation methods for the characterization of the texture of mass lesions and architectural distortion. We applied the concept of lacunarity to the description of the spatial distribution of the pixel intensities in mammographic images. These methods were tested with a set of 57 breast masses and 60 normal breast parenchyma (dataset1), and with another set of 19 architectural distortions and 41 normal breast parenchyma (dataset2). Support vector machines (SVM) were used as a pattern classification method for tumor classification. Results: Experimental results showed that the fractal dimension of region of interest (ROIs) depicting mass lesions and architectural distortion was statistically significantly lower than that of normal breast parenchyma for all five methods. Receiver operating characteristic (ROC) analysis showed that fractional Brownian motion (FBM) method generated the highest area under ROC curve (A z = 0.839 for dataset1, 0.828 for dataset2, respectively) among five methods for both datasets. Lacunarity analysis showed that the ROIs depicting mass lesions and architectural distortion had higher lacunarities than those of ROIs depicting normal breast parenchyma. The combination of FBM fractal dimension and lacunarity yielded the highest A z value (0.903 and 0.875, respectively) than those based on single feature alone for both given datasets. The application of the SVM improved the performance of the fractal-based features in differentiating tumor lesions from normal breast parenchyma by generating higher A z value. Conclusion: FBM texture model is the most appropriate model for characterizing mammographic images due to self-affinity assumption of the method being a better approximation. Lacunarity is an effective counterpart measure of the fractal dimension in texture feature extraction in mammographic images. The classification results obtained in this work suggest that the SVM is an effective method with great potential for classification in mammographic image analysis.

Interactions between age and apoE genotype on fasting and postprandial triglycerides levels

Objective The influences of genetic determinants on the magnitude of postprandial lipaemia are presently unclear. Here the impact of the common apolipoprotein (apo)E epsilon mutation on the postprandial triglyceride (TG) response is determined, along with an assessment of genotype penetrance according to age, body mass index and gender. Methods and results Healthy adults (n = 251) underwent a postprandial investigation, in which blood samples were taken at regular intervals after a test breakfast (0 min, 49 g fat) and lunch (330 min, 29 g fat) until 480 min after the test breakfast. There was a significant impact of apoE genotype on fasting total cholesterol (TC), (P = 0.027), LDL-cholesterol (LDL-C), (P = 0.008), and %LDL3 (P = 0.001), with higher and lower levels in the E4 and E2 carriers respectively relative to the E3/E3 genotype. Reflective of a higher fasting TG (P = 0.001), a significantly higher area under the curve for the postprandial TG response (TG AUC) was evident in the E4 carriers relative to the E3/E3 group (P = 0.038). In the group as a whole, a significant age × genotype interaction was observed for fasting TC (P = 0.021). In the participants >50 years there was a significant impact of genotype on TC (P = 0.005), LDL-C (P = 0.001) and TAG AUC (P = 0.028). Conclusions It is possible that an exaggerated postprandial lipaemia contributes to the increased coronary heart disease risk associated with carriers of the E4 allele; an effect which is more evident in older adults.