A dynamic sampling methodology for plasma etch processes using Gaussian process regression

Plasma etch is a key process in modern semiconductor manufacturing facilities as it offers process simplification and yet greater dimensional tolerances compared to wet chemical etch technology. The main challenge of operating plasma etchers is to maintain a consistent etch rate spatially and temporally for a given wafer and for successive wafers processed in the same etch tool. Etch rate measurements require expensive metrology steps and therefore in general only limited sampling is performed. Furthermore, the results of measurements are not accessible in real-time, limiting the options for run-to-run control. This paper investigates a Virtual Metrology (VM) enabled Dynamic Sampling (DS) methodology as an alternative paradigm for balancing the need to reduce costly metrology with the need to measure more frequently and in a timely fashion to enable wafer-to-wafer control. Using a Gaussian Process Regression (GPR) VM model for etch rate estimation of a plasma etch process, the proposed dynamic sampling methodology is demonstrated and evaluated for a number of different predictive dynamic sampling rules. © 2013 IEEE.

Efficient Marginal Likelihood Computation for Gaussian Process Regression

In a Bayesian learning setting, the posterior distribution of a predictive model arises from a trade-off between its prior distribution and the conditional likelihood of observed data. Such distribution functions usually rely on additional hyperparameters which need to be tuned in order to achieve optimum predictive performance; this operation can be efficiently performed in an Empirical Bayes fashion by maximizing the posterior marginal likelihood of the observed data. Since the score function of this optimization problem is in general characterized by the presence of local optima, it is necessary to resort to global optimization strategies, which require a large number of function evaluations. Given that the evaluation is usually computationally intensive and badly scaled with respect to the dataset size, the maximum number of observations that can be treated simultaneously is quite limited. In this paper, we consider the case of hyperparameter tuning in Gaussian process regression. A straightforward implementation of the posterior log-likelihood for this model requires O(N^3) operations for every iteration of the optimization procedure, where N is the number of examples in the input dataset. We derive a novel set of identities that allow, after an initial overhead of O(N^3), the evaluation of the score function, as well as the Jacobian and Hessian matrices, in O(N) operations. We prove how the proposed identities, that follow from the eigendecomposition of the kernel matrix, yield a reduction of several orders of magnitude in the computation time for the hyperparameter optimization problem. Notably, the proposed solution provides computational advantages even with respect to state of the art approximations that rely on sparse kernel matrices.

An Adaptive Machine Learning Decision System for Flexible Predictive Maintenance

Process monitoring and Predictive Maintenance (PdM) are gaining increasing attention in most manufacturing environments as a means of reducing maintenance related costs and downtime. This is especially true in industries that are data intensive such as semiconductor manufacturing. In this paper an adaptive PdM based flexible maintenance scheduling decision support system, which pays particular attention to associated opportunity and risk costs, is presented. The proposed system, which employs Machine Learning and regularized regression methods, exploits new information as it becomes available from newly processed components to refine remaining useful life estimates and associated costs and risks. The system has been validated on a real industrial dataset related to an Ion Beam Etching process for semiconductor manufacturing.

On Regression Methods for Virtual Metrology in Semiconductor Manufacturing

Virtual metrology (VM) aims to predict metrology values using sensor data from production equipment and physical metrology values of preceding samples. VM is a promising technology for the semiconductor manufacturing industry as it can reduce the frequency of in-line metrology operations and provide supportive information for other operations such as fault detection, predictive maintenance and run-to-run control. The prediction models for VM can be from a large variety of linear and nonlinear regression methods and the selection of a proper regression method for a specific VM problem is not straightforward, especially when the candidate predictor set is of high dimension, correlated and noisy. Using process data from a benchmark semiconductor manufacturing process, this paper evaluates the performance of four typical regression methods for VM: multiple linear regression (MLR), least absolute shrinkage and selection operator (LASSO), neural networks (NN) and Gaussian process regression (GPR). It is observed that GPR performs the best among the four methods and that, remarkably, the performance of linear regression approaches that of GPR as the subset of selected input variables is increased. The observed competitiveness of high-dimensional linear regression models, which does not hold true in general, is explained in the context of extreme learning machines and functional link neural networks.

Which biomarkers are predictive specifically for cardiovascular or for non-cardiovascular mortality in men? Evidence from the Caerphilly Prospective Study (CaPS)

OBJECTIVE: To examine a panel of 28 biomarkers for prediction of cardiovascular disease (CVD) and non-CVD mortality in a population-based cohort of men.

METHODS: Starting in 1979, middle-aged men in Caerphilly underwent detailed medical examination. Subsequently 2171 men were re-examined during 1989-1993, and fasting blood samples obtained from 1911 men (88%). Fibrinogen, viscosity and white cell count (WCC), routine biochemistry tests and lipids were analysed using fresh samples. Stored aliquots were later analysed for novel biomarkers. Statistical analysis of CVD and non-CVD mortality follow-up used competing risk Cox regression models with biomarkers in thirds tested at the 1% significance level after covariate adjustment.

RESULTS: During an average of 15.4years follow-up, troponin (subhazard ratio per third 1.71, 95% CI 1.46-1.99) and B-natriuretic peptide (BNP) (subhazard ratio per third 1.54, 95% CI 1.34-1.78) showed strong trends with CVD death but not with non-CVD death. WCC and fibrinogen showed similar weaker findings. Plasma viscosity, growth differentiation factor 15 (GDF-15) and interleukin-6 (IL-6) were associated positively with both CVD death and non-CVD death while total cholesterol was associated positively with CVD death but negatively with non-CVD death. C-reactive protein (C-RP), alkaline phosphatase, gamma-glutamyltransferase (GGT), retinol binding protein 4 (RBP-4) and vitamin B6 were significantly associated only with non-CVD death, the last two negatively. Troponin, BNP and IL-6 showed evidence of diminishing associations with CVD mortality through follow-up.

CONCLUSION: Biomarkers for cardiac necrosis were strong, specific predictors of CVD mortality while many inflammatory markers were equally predictive of non-CVD mortality.

Supervised Aggregative Feature Extraction for Big Data Time Series Regression

In many applications, and especially those where batch processes are involved, a target scalar output of interest is often dependent on one or more time series of data. With the exponential growth in data logging in modern industries such time series are increasingly available for statistical modeling in soft sensing applications. In order to exploit time series data for predictive modelling, it is necessary to summarise the information they contain as a set of features to use as model regressors. Typically this is done in an unsupervised fashion using simple techniques such as computing statistical moments, principal components or wavelet decompositions, often leading to significant information loss and hence suboptimal predictive models. In this paper, a functional learning paradigm is exploited in a supervised fashion to derive continuous, smooth estimates of time series data (yielding aggregated local information), while simultaneously estimating a continuous shape function yielding optimal predictions. The proposed Supervised Aggregative Feature Extraction (SAFE) methodology can be extended to support nonlinear predictive models by embedding the functional learning framework in a Reproducing Kernel Hilbert Spaces setting. SAFE has a number of attractive features including closed form solution and the ability to explicitly incorporate first and second order derivative information. Using simulation studies and a practical semiconductor manufacturing case study we highlight the strengths of the new methodology with respect to standard unsupervised feature extraction approaches.

Natural Killer Cell-like signature observed in locally advanced rectal cancer after neoadjuvant chemoradiotherapy in a tumour regression grade dependent manner

Background: Around 10-15% of patients with locally advanced rectal cancer (LARC) undergo a pathologically complete response (TRG4) to neoadjuvant chemoradiotherapy; the rest of patients exhibit a spectrum of tumour regression (TRG1-3). Understanding therapy-related genomic alterations may help us to identify underlying biology or novel targets associated with response that could increase the efficacy of therapy in patients that do not benefit from the current standard of care.
Methods: 48 FFPE rectal cancer biopsies and matched resections were analysed using the WG-DASL HumanHT-12_v4 Beadchip array on the illumina iScan. Bioinformatic analysis was conducted in Partek genomics suite and R studio. Limma and glmnet packages were used to identify genes differentially expressed between tumour regression grades. Validation of microarray results will be carried out using IHC, RNAscope and RT-PCR.
Results: Immune response genes were observed from supervised analysis of the biopsies which may have predictive value. Differential gene expression from the resections as well as pre and post therapy analysis revealed induction of genes in a tumour regression dependent manner. Pathway mapping and Gene Ontology analysis of these genes suggested antigen processing and natural killer mediated cytotoxicity respectively. The natural killer-like gene signature was switched off in non-responders and on in the responders. IHC has confirmed the presence of Natural killer cells through CD56+ staining.
Conclusion: Identification of NK cell genes and CD56+ cells in patients responding to neoadjuvant chemoradiotherapy warrants further investigation into their association with tumour regression grade in LARC. NK cells are known to lyse malignant cells and determining whether their presence is a cause or consequence of response is crucial. Interrogation of the cytokines upregulated in our NK-like signature will help guide future in vitro models.

Preoperative radial artery volume flow is predictive of arteriovenous fistula outcomes

Objective: Guidelines recommend the creation of a wrist radiocephalic arteriovenous fistula (RAVF) as initial hemodialysis vascular access. This study explored the potential of preoperative ultrasound vessel measurements to predict AVF failure to mature (FTM) in a cohort of patients with end-stage renal disease in Northern Ireland

.Methods: A retrospective analysis was performed of all patients who had preoperative ultrasound mapping of upper limb blood vessels carried out from August 2011 to December 2014 and whose AVF reached a functional outcome by March 2015.

Results: There were 152 patients (97% white) who had ultrasound mapping andan AVF functional outcome recorded; 80 (54%) had an upper arm AVF created, and 69 (46%) had a RAVF formed. Logistic regression revealed that female gender (odds ratio [OR], 2.5; 95% confidence interval [CI], 1.12-5.55; P = .025), minimum venous diameter (OR, 0.6; 95% CI, 0.39-0.95; P = .029), and RAVF (OR, 0.4; 95% CI, 0.18-0.89; P = .026) were associated with FTM. On subgroup analysis of the RAVF group, RAVFs with an arterial volume flow <50 mL/min were seven times as likely to fail as RAVFs with higher volume flows (OR, 7.0; 95% CI, 2.35-20.87; P < .001).

Conclusions: In this cohort, a radial artery flow rate <50 mL/min was associated with a sevenfold increased risk of FTM in RAVF, which to our knowledge has not been previously reported in the literature. Preoperative ultrasound mapping adds objective assessment in the clinical prediction of AVF FTM.

An Experimental and Predictive Evaluation of Unsteady Gas Flow through Automotive Catalyst Elements

The incorporation of one-dimensional simulation codes within engine modelling applications has proved to be a useful tool in evaluating unsteady gas flow through elements in the exhaust system. This paper reports on an experimental and theoretical investigation into the behaviour of unsteady gas flow through catalyst substrate elements. A one-dimensional (1-D) catalyst model has been incorporated into a 1-D simulation code to predict this behaviour.

Experimental data was acquired using a ‘single pulse’ test rig. Substrate samples were tested under ambient conditions in order to investigate a range of regimes experienced by the catalyst during operation. This allowed reflection and transmission characteristics to be quantified in relation to both geometric and physical properties of substrate elements. Correlation between measured and predicted results is demonstrably good and the model provides an effective analysis tool for evaluating unsteady gas flow through different catalytic converter designs.

Can diagnostic triage by general practitioners or rheumatology nurses improve the positive predictive value of referrals to early arthritis clinics?

Objectives: To determine whether diagnostic triage by general practitioners (GPs) or rheumatology nurses (RNs) can improve the positive predictive value of referrals to early arthritis clinics (EACs).

Methods: Four GPs and two RNs were trained in the assessment of early in?ammatory arthritis (IA) by four visits to an EAC supervised by hospital rheumatologists. Patients referred to one of three EACs were recruited for study and assessed independently by a GP, an RN and one of six rheumatologists. Each assessor was asked to record their clinical ?ndings and whether they considered the patient to have IA. Each was then asked to judge the appropriateness of the referral according to predetermined guidelines. The rheumatologists had been shown previously to have a satisfactory level of agreement in the assessment of IA.

Results: Ninety-six patients were approached and all consented to take part in the study. In 49 cases (51%), the rheumatologist judged that the patient had IA and that the referral was appropriate. The assessments of GPs and RNs were compared with those of the rheumatologists. Levels of agreement were measured using the kappa value, where 1.0 represents total unanimity. The kappa value was
0.77 for the GPs when compared with the rheumatologists and 0.79 for the RNs. Signi?cant stiffness in the morning or after rest and objective joint swelling were the most important clinical features enabling the GPs and RNs to discriminate between IA and non-IA conditions.

Conclusion: Diagnostic triage by GPs or RNs improved the positive predictive value of referrals to an EAC with a degree of accuracy approaching that of a group of experienced rheumatologists.

Relationship between biomedical catheter surface properties and lubricity as determined using textural analysis and multiple regression analysis

In this study, the surface properties of and work required to remove 12 commercially available and developmental catheters from a model biological medium (agar), a measure of catheter lubricity, were characterised and the relationships between these properties were examined using multiple regression and correlation analysis. The work required for removal of catheter sections (7 cm) from a model biological medium (1% w/w agar) were examined using tensile analysis. The water wettability of the catheters were characterised using dynamic contact angle analysis, whereas surface roughness was determined using atomic force microscopy. Significant differences in the ease of removal were observed between the various catheters, with the silicone-based materials generally exhibiting the greatest ease of removal. Similarly, the catheters exhibited a range of advancing and receding contact angles that were dependent on the chemical nature of each catheter. Finally, whilst the microrugosities of the various catheters differed, no specific relationship to the chemical nature of the biomaterial was apparent. Using multiple regression analysis, the relationship between ease of removal, receding contact angle and surface roughness was defined as: Work done (N mm) 17.18 + 0.055 Rugosity (nm)-0.52 Receding contact angle (degrees) (r = 0.49). Interestingly, whilst the relationship between ease of removal and surface roughness was significant (r = 0.48, p = 0.0005), in which catheter lubricity increased as the surface roughness decreased, this was not the case with the relationship between ease of removal and receding contact angle (r = -0.18, p > 0.05). This study has therefore uniquely defined the contributions of each of these surface properties to catheter lubricity. Accordingly, in the design of urethral catheters. it is recommended that due consideration should be directed towards biomaterial surface roughness to ensure maximal ease of catheter removal. Furthermore, using the method described in this study, differences in the lubricity of the various catheters were observed that may be apparent in their clinical use. (C) 2003 Elsevier Ltd. All rights reserved.

Do Total and High Density Lipoprotein Cholesterol and Triglycerides Act Independently in the Prediction of Ischemic Heart Disease? Ten-Year Follow-Up of Caerphilly and Speedwell Cohorts

Several studies have suggested that men with raised plasma triglycerides (TGs) in combination with adverse levels of other lipids may be at special risk of subsequent ischemic heart disease (IHD). We examined the independent and combined effects of plasma lipids at 10 years of follow-up. We measured fasting TGs, total cholesterol (TC), and high density lipoprotein cholesterol (HDLC) in 4362 men (aged 45 to 63 years) from 2 study populations and reexamined them at intervals during a 10-year follow-up. Major IHD events (death from IHD, clinical myocardial infarction, or ECG-defined myocardial infarction) were recorded. Five hundred thirty-three major IHD events occurred. All 3 lipids were strongly and independently predictive of IHD after 10 years of follow-up. Subjects were then divided into 27 groups (ie, 33) by the tertiles of TGs, TC, and HDLC. The number of events observed in each group was compared with that predicted by a logistic regression model, which included terms for the 3 lipids (without interactions) and potential confounding variables. The incidence of IHD was 22.6% in the group with the lipid risk factor combination with the highest expected risk (high TGs, high TC, and low HDLC) and 4.7% in the group with the lowest expected risk (P