Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Background The SARS-CoV-2 Delta variant (B.1.617.2), first detected in India, has rapidly become the dominant variant in England. Early reports suggest this variant has an increased growth rate suggesting increased transmissibility. This study indirectly assessed differences in transmissibility between the emergent Delta variant compared to the previously dominant Alpha variant (B.1.1.7). Methods A matched case-control study was conducted to estimate the odds of household transmission (≥ 2 cases within 14 days) for Delta variant index cases compared with Alpha cases. Cases were derived from national surveillance data (March to June 2021). One-to-two matching was undertaken on geographical location of residence, time period of testing and property type, and a multivariable conditional logistic regression model was used for analysis. Findings In total 5,976 genomically sequenced index cases in household clusters were matched to 11,952 sporadic index cases (single case within a household). 43.3% (n=2,586) of cases in household clusters were confirmed Delta variant compared to 40.4% (n= 4,824) of sporadic cases. The odds ratio of household transmission was 1.70 among Delta variant cases (95% CI 1.48-1.95, p <0.001) compared to Alpha cases after adjusting for age, sex, ethnicity, index of multiple deprivation (IMD), number of household contacts and vaccination status of index case. Interpretation We found evidence of increased household transmission of SARS-CoV-2 Delta variant, potentially explaining its success at displacing Alpha variant as the dominant strain in England. With the Delta variant now having been detected in many countries worldwide, the understanding of the transmissibility of this variant is important for informing infection prevention and control policies internationally.

Household transmission of COVID-19 cases associated with SARS-CoV-2 delta variant (B.1.617.2): national case-control study

The SARS-CoV-2 Delta variant (B.1.617.2), first detected in India, has rapidly become the dominant variant in England. Early reports suggest this variant has an increased growth rate suggesting increased transmissibility. This study indirectly assessed differences in transmissibility between the emergent Delta variant compared to the previously dominant Alpha variant (B.1.1.7).A matched case-control study was conducted to estimate the odds of household transmission (≥ 2 cases within 14 days) for Delta variant index cases compared with Alpha cases. Cases were derived from national surveillance data (March to June 2021). One-to-two matching was undertaken on geographical location of residence, time period of testing and property type, and a multivariable conditional logistic regression model was used for analysis.In total 5,976 genomically sequenced index cases in household clusters were matched to 11,952 sporadic index cases (single case within a household). 43.3% (n=2,586) of cases in household clusters were confirmed Delta variant compared to 40.4% (n= 4,824) of sporadic cases. The odds ratio of household transmission was 1.70 among Delta variant cases (95% CI 1.48-1.95, p <0.001) compared to Alpha cases after adjusting for age, sex, ethnicity, index of multiple deprivation (IMD), number of household contacts and vaccination status of index case.We found evidence of increased household transmission of SARS-CoV-2 Delta variant, potentially explaining its success at displacing Alpha variant as the dominant strain in England. With the Delta variant now having been detected in many countries worldwide, the understanding of the transmissibility of this variant is important for informing infection prevention and control policies internationally. 

Industrial processes with multiple operating grades have become increasingly important in satisfying the requirements of agile manufacturing and a diversified market. However, because of the unknow...

Development of Adversarial Transfer Learning Soft Sensor for Multigrade Processes

Integrating adaptive moving window and just-in-time learning paradigms for soft-sensor design

Coronavirus disease 2019 (COVID-19) has emerged in December 2019 when the first case was reported in Wuhan, China and turned into a pandemic with 27 million (September 9th) cases. Currently, there are over 95,000 complete genome sequences of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus causing COVID-19, in public databases, accompanying a growing number of studies. Nevertheless, there is still much to learn about the viral population variation when the virus is evolving as it continues to spread. We have analyzed SARS-CoV-2 genomes to identify the most variant sites, as well as the stable, conserved ones in samples collected in the Netherlands until June 2020. We identified the most frequent mutations in different geographies. We also performed a phylogenetic study focused on the Netherlands to detect novel variants emerging in the late stages of the pandemic and forming local clusters. We investigated the S and N proteins on SARS-CoV-2 genomes in the Netherlands and found the most variant and stable sites to guide development of diagnostics assays and vaccines. We observed that while the SARS-CoV-2 genome has accumulated mutations, diverging from reference sequence, the variation landscape is dominated by four mutations globally, suggesting the current reference does not represent the virus samples circulating currently. In addition, we detected novel variants of SARS-CoV-2 almost unique to the Netherlands that form localized clusters and region-specific sub-populations indicating community spread. We explored SARS-CoV-2 variants in the Netherlands until June 2020 within a global context; our results provide insight into the viral population diversity for localized efforts in tracking the transmission of COVID-19, as well as sequenced-based approaches in diagnostics and therapeutics. We emphasize that little diversity is observed globally in recent samples despite the increased number of mutations relative to the established reference sequence. We suggest sequence-based analyses should opt for a consensus representation to adequately cover the genomic variation observed to speed up diagnostics and vaccine design.

/pdf/emergence-of-novel-sars-cov-2-variants-in-the-netherlands-362dui8v3f.pdf

Emergence of novel SARS-CoV-2 variants in the Netherlands.

In this study, we analyzed SARS-CoV-2 genomes in the Netherlands, in the context of global viral population since the beginning of the pandemic. We have identified the most variant sites on the whole genome as well as the stable, conserved ones on the S and N proteins. We found four mutations, S:D614G, NSP12b:P314L, NSP3:F106F, to be the most frequent ones that dominate the SARS-CoV-2 population outside of China. We detected novel variants of SARS-CoV-2 almost unique to the Netherlands that form localized clusters, indicating community spread. We emphasize that while SARS-CoV-2 is evolving, and the number of mutations from the reference sequence is increasing, we observe only little diversity in the new variants as we enter the later stages of the pandemic. Our analyses suggest we have diverged away from the current SARS-CoV-2 reference enough that the reference should be re-evaluated to represent the current viral population more accurately. We assert our work provides valuable information on the genetic diversity of SARS-CoV-2 and its local dynamics in the Netherlands, especially for DNA-based diagnostic, therapeutic or vaccine development against COVID-19. We suggest sequence-based analyses should opt for a consensus representation to adequately cover the genomic variation observed.

/pdf/emergence-of-novel-sars-cov-2-variants-in-the-netherlands-2h9zzr6rj2.pdf

Emergence of Novel SARS-CoV-2 Variants in the Netherlands

Data-driven soft-sensors are statistical models constructed from historical data, and expected to perform well both in normal and novel process conditions. Numerous adaptive mechanisms for soft-sensors have been developed, but more work is required to develop appropriate statistical modeling tools for chemical processes, which yield highly collinear measurements in changing operating conditions. In the current study, numerous predictor subset selection and statistical learning tools are employed on 12 different data sets from a crude distillation unit simulated in SimSci PROII™, and the generalization capacities of the models are evaluated using the test points inside and outside the operating regions of the training sets. Relevance vector machine is found to be promising for adaptive modeling, with accuracies close to those of neural network models.

Soft-Sensor Design for a Crude Distillation Unit Using Statistical Learning Methods

The success of antibiotics as a therapeutic agent has led to their ineffectiveness. The continuous use and misuse in clinical and non-clinical areas have led to the emergence and spread of antibiotic-resistant bacteria and its genetic determinants. This is a multi-dimensional problem that has now become a global health crisis. Antibiotic resistance research has primarily focused on the clinical healthcare sectors while overlooking the non-clinical sectors. The increasing antibiotic usage in the environment – including animals, plants, soil, and water – are drivers of antibiotic resistance and function as a transmission route for antibiotic resistant pathogens and is a source for resistance genes. These natural compartments are interconnected with each other and humans, allowing the spread of antibiotic resistance via horizontal gene transfer between commensal and pathogenic bacteria. Identifying and understanding genetic exchange within and between natural compartments can provide insight into the transmission, dissemination, and emergence mechanisms. The development of high-throughput DNA sequencing technologies has made antibiotic resistance research more accessible and feasible. In particular, the combination of metagenomics and powerful bioinformatic tools and platforms have facilitated the identification of microbial communities and has allowed access to genomic data by bypassing the need for isolating and culturing microorganisms. This review aimed to reflect on the different sequencing techniques, metagenomic approaches, and bioinformatics tools and pipelines with their respective advantages and limitations for antibiotic resistance research. These approaches can provide insight into resistance mechanisms, the microbial population, emerging pathogens, resistance genes, and their dissemination. This information can influence policies, develop preventative measures and alleviate the burden caused by antibiotic resistance.

Aysun Urhan

Papers

Integrating adaptive moving window and just-in-time learning paradigms for soft-sensor design

Emergence of novel SARS-CoV-2 variants in the Netherlands.

Emergence of Novel SARS-CoV-2 Variants in the Netherlands

Soft-Sensor Design for a Crude Distillation Unit Using Statistical Learning Methods

Metagenomic-based surveillance systems for antibiotic resistance in non-clinical settings