HLA-B*38:02:01 predicts carbimazole/methimazole-induced agranulocytosis.

Thioamides antithyroid-drugs (ATDs) are important in hyperthyroid disease management. Identification of the susceptibility locus of ATD-induced agranulocytosis is important for clinical management. We performed a genome-wide association study (GWAS) involving 20 patients with ATD-induced agranulocytosis and 775 healthy controls. The top finding was further replicated. A single-nucleotide polymorphism (SNP), rs185386680, showed the strongest association with ATD-induced agranulocytosis in GWAS (odds ratio (OR) = 36.4; 95% confidence interval (CI) = 12.8-103.7; P = 1.3 × 10(-24)) and replication (OR = 37; 95% CI = 3.7-367.4; P = 9.6 × 10(-7)). HLA-B*38:02:01 was in complete linkage disequilibrium with rs185386680. High-resolution HLA typing confirmed that HLA-B*38:02:01 was associated with carbimazole (CMZ)/methimazole (MMI)-induced agranulocytosis (OR = 265.5; 95% CI = 27.9-2528.0; P = 2.5 × 10(-14)), but not associated with propylthiouracil (PTU). The positive and negative predictive values of HLA-B*38:02:01 in predicting CMZ/MMI-induced agranulocytosis were 0.07 and 0.999. Approximately 211 cases need to be screened to prevent one case. Screening for the risk allele will be useful in preventing agranulocytosis in populations in which the frequency of the risk allele is high.

A comparative study on kernel-based probabilistic neural networks for speaker verification.

This paper compares kernel-based probabilistic neural networks for speaker verification based on 138 speakers of the YOHO corpus. Experimental evaluations using probabilistic decision-based neural networks (PDBNNs), Gaussian mixture models (GMMs) and elliptical basis function networks (EBFNs) as speaker models were conducted. The original training algorithm of PDBNNs was also modified to make PDBNNs appropriate for speaker verification. Results show that the equal error rate obtained by PDBNNs and GMMs is less than that of EBFNs (0.33% vs. 0.48%), suggesting that GMM- and PDBNN-based speaker models outperform the EBFN ones. This work also finds that the globally supervised learning of PDBNNs is able to find decision thresholds that not only maintain the false acceptance rates to a low level but also reduce their variation, whereas the ad-hoc threshold-determination approach used by the EBFNs and GMMs causes a large variation in the error rates. This property makes the performance of PDBNN-based systems more predictable.

Estimation of elliptical basis function parameters by the EM algorithm with application to speaker verification.

This paper proposes to incorporate full covariance matrices into the radial basis function (RBF) networks and to use the expectation-maximization (EM) algorithm to estimate the basis function parameters. The resulting networks, referred to as elliptical basis function (EBF) networks, are evaluated through a series of text-independent speaker verification experiments involving 258 speakers from a phonetically balanced, continuous speech corpus (TIMIT).We propose a verification procedure using RBF and EBF networks as speaker models and show that the networks are readily applicable to verifying speakers using LP-derived cepstral coefficients as features. Experimental results show that small EBF networks with basis function parameters estimated by the EM algorithm outperform the large RBF networks trained in the conventional approach. The results also show that the equal error rate achieved by the EBF networks is about two-third of that achieved by the vetor quantization (VQ)-based speaker models.

Intracranial delivery of recombinant nerve growth factor: release kinetics and protein distribution for three delivery systems.

PURPOSE: Three different polymeric delivery systems, composed of either poly(ethylene-co-vinyl acetate) (EVAc) or poly(lactide-co-glycolide) (PLGA), were used to administer recombinant human nerve growth factor (rhNGF) intracranially in rats. METHODS: The delivery systems were characterized with respect to release kinetics, both in the brain and in well-stirred buffer solutions. RESULTS: During incubation in buffered saline, the delivery systems released rhNGF in distinct patterns: sustained (EVAc), immediate (PLGA1) and delayed (PLGA2). One 10-mg delivery system was implanted in each rat and an ELISA technique was used to determine the amount of rhNGF in 1-mm coronal brain slices produced immediately after removal of the delivery system. High levels of rhNGF (as high as 60,000 ng in a brain slice of approximately 50 microliters) were recovered from the brain tissue at 1, 2, and 4 weeks after implantation. With all three delivery systems, the amount of rhNGF in each brain slice decreased exponentially with distance from the implant site: the distance over which concentration decreased by 10-fold was 2-3 mm for all delivery systems. When rhNGF release was moderate (10 to 200 ng rhNGF/day), the total amount of rhNGF in the brain increased linearly with release rate, suggesting an overall rate of rhNGF elimination of 0.4 hr-1 or a half-life of 1.7 hr. With higher release rates (500 to 50,000 ng rhNGF/day), total amounts of rhNGF in the brain were considerably higher than anticipated based on this rate of elimination. CONCLUSIONS: Polymeric controlled release can provide high, localized doses of rhNGF in the brain. All of the experimental data were consistent with penetration of rhNGF through the brain tissue with a diffusion coefficient approximately 8 x 10(-7) cm2/s, which is approximately 50% of the diffusion coefficient in water.

Adding learning to cellular genetic algorithms for training recurrent neural networks.

This paper proposes a hybrid optimization algorithm which combines the efforts of local search (individual learning) and cellular genetic algorithms (GA's) for training recurrent neural networks (RNN's). Each weight of an RNN is encoded as a floating point number, and a concatenation of the numbers forms a chromosome. Reproduction takes place locally in a square grid with each grid point representing a chromosome. Two approaches, Lamarckian and Baldwinian mechanisms, for combining cellular GA's and learning have been compared. Different hill-climbing algorithms are incorporated into the cellular GA's as learning methods. These include the real-time recurrent learning (RTRL) and its simplified versions, and the delta rule. The RTRL algorithm has been successively simplified by freezing some of the weights to form simplified versions. The delta rule, which is the simplest form of learning, has been implemented by considering the RNN's as feedforward networks during learning. The hybrid algorithms are used to train the RNN's to solve a long-term dependency problem. The results show that Baldwinian learning is inefficient in assisting the cellular GA. It is conjectured that the more difficult it is for genetic operations to produce the genotypic changes that match the phenotypic changes due to learning, the poorer is the convergence of Baldwinian learning. Most of the combinations using the Lamarckian mechanism show an improvement in reducing the number of generations required for an optimum network; however, only a few can reduce the actual time taken. Embedding the delta rule in the cellular GA's has been found to be the fastest method. It is also concluded that learning should not be too extensive if the hybrid algorithm is to be benefit from learning.