Improved Nonlinear PCA for Process Monitoring Using Support Vector Data Description

Nonlinear principal component analysis (PCA) based on neural networks has drawn significant attention as a monitoring tool for complex nonlinear processes, but there remains a difficulty with determining the optimal network topology. This paper exploits the advantages of the Fast Recursive Algorithm, where the number of nodes, the location of centres, and the weights between the hidden layer and the output layer can be identified simultaneously for the radial basis function (RBF) networks. The topology problem for the nonlinear PCA based on neural networks can thus be solved. Another problem with nonlinear PCA is that the derived nonlinear scores may not be statistically independent or follow a simple parametric distribution. This hinders its applications in process monitoring since the simplicity of applying predetermined probability distribution functions is lost. This paper proposes the use of a support vector data description and shows that transforming the nonlinear principal components into a feature space allows a simple statistical inference. Results from both simulated and industrial data confirm the efficacy of the proposed method for solving nonlinear principal component problems, compared with linear PCA and kernel PCA.

Development and clinical validation of multiplex TaqMan® assays for rapid diagnosis of viral gastroenteritis.

There is a need to provide rapid, sensitive, and often high throughput detection of pathogens in diagnostic virology. Viral gastroenteritis is a serious health issue often leading to hospitalization in the young, the immunocompromised and the elderly. The common causes of viral gastroenteritis include rotavirus, norovirus (genogroups I and II), astrovirus, and group F adenoviruses (serotypes 40 and 41). This article describes the work-up of two internally controlled multiplex, probe-based PCR assays and reports on the clinical validation over a 3-year period, March 2007 to February 2010. Multiplex assays were developed using a combination of TaqMan™ and minor groove binder (MGB™) hydrolysis probes. The assays were validated using a panel of 137 specimens, previously positive via a nested gel-based assay. The assays had improved sensitivity for adenovirus, rotavirus, and norovirus (97.3% vs. 86.1%, 100% vs. 87.8%, and 95.1% vs. 79.5%, respectively) and also more specific for targets adenovirus, rotavirus, and norovirus (99% vs. 95.2%, 100% vs. 93.6%, and 97.9% vs. 92.3%, respectively). For the specimens tested, both assays had equal sensitivity and specificity for astrovirus (100%). Overall the probe-based assays detected 16 more positive specimens than the nested gel-based assay. Post-introduction to the routine diagnostic service, a total of 9,846 specimens were processed with multiplex 1 and 2 (7,053 pediatric, 2,793 adult) over the 3-year study period. This clinically validated, probe-based multiplex testing algorithm allows highly sensitive and timely diagnosis of the four most prominent causes of viral gastroenteritis.