Evaluating sampling strategies and logistic regression methods for modelling complex land cover changes

The role of land cover change as a significant component of global change has become increasingly recognized in recent decades. Large databases measuring land cover change, and the data which can potentially be used to explain the observed changes, are also becoming more commonly available. When developing statistical models to investigate observed changes, it is important to be aware that the chosen sampling strategy and modelling techniques can influence results. We present a comparison of three sampling strategies and two forms of grouped logistic regression models (multinomial and ordinal) in the investigation of patterns of successional change after agricultural land abandonment in Switzerland. Results indicated that both ordinal and nominal transitional change occurs in the landscape and that the use of different sampling regimes and modelling techniques as investigative tools yield different results. Synthesis and applications. Our multimodel inference identified successfully a set of consistently selected indicators of land cover change, which can be used to predict further change, including annual average temperature, the number of already overgrown neighbouring areas of land and distance to historically destructive avalanche sites. This allows for more reliable decision making and planning with respect to landscape management. Although both model approaches gave similar results, ordinal regression yielded more parsimonious models that identified the important predictors of land cover change more efficiently. Thus, this approach is favourable where land cover change pattern can be interpreted as an ordinal process. Otherwise, multinomial logistic regression is a viable alternative.

The Change Questionnaire predicts change in hazardous tobacco and alcohol use.

INTRODUCTION: Assessing motivation for change is deemed an important step in the treatment process that allows further refinement of the intervention in motivational interviewing (MI) and brief MI (BMI) adaptations. During MI (and BMI) sessions, motivation for change is expressed by the client as "change talk", i.e. all statements inclined toward or away from change. We tested the predictive validity of the Change Questionnaire, a 12-item instrument assessing motivation to change, on hazardous tobacco and alcohol use. METHODS: As part of the baseline measurements for a randomized controlled trial on multi-substance BMI at the Lausanne recruitment center (army conscription is mandatory in Switzerland for males at age 20, and thus provides a unique opportunity to address a non-clinical and largely representative sample of young men), 213 participants completed the questionnaire on tobacco and 95 on alcohol and were followed-up six months later. The overall Change Questionnaire score and its six subscales (Desire, Ability, Reasons, Need, Commitment, and Taking steps) were used as predictors of hazardous tobacco use (defined as daily smoking) and hazardous alcohol use (defined as more than one occasion with six standard drinks or more per month, and/or more than 21 standard drinks per week) in bivariate logistic regression models at follow-up. RESULTS: Higher overall Change scores were significant predictors of decreased risk for hazardous tobacco (odds ratio [OR] = 0.83, p = 0.046) and alcohol (OR = 0.76, p = 0.03) use. Several sub-dimensions were associated with the outcomes in bivariate analyses. Using a principal components analysis to reduce the number of predictors for multivariate models, we obtained two components. 'Ability to change' was strongly related to change in hazardous tobacco use (OR = 0.54, p < 0.001), the second we interpreted as 'Other change language dimensions' and which was significantly related to change in hazardous alcohol use (OR = 0.81, p = 0.05). CONCLUSIONS: The present findings lend initial support for the predictive validity of the Change Questionnaire on hazardous tobacco and alcohol use, making it an interesting and potentially useful tool for assessing motivation to change among young males.

Bayesian Model Averaging in the Instrumental Variable Regression Model

This paper considers the instrumental variable regression model when there is uncertainty about the set of instruments, exogeneity restrictions, the validity of identifying restrictions and the set of exogenous regressors. This uncertainty can result in a huge number of models. To avoid statistical problems associated with standard model selection procedures, we develop a reversible jump Markov chain Monte Carlo algorithm that allows us to do Bayesian model averaging. The algorithm is very exible and can be easily adapted to analyze any of the di¤erent priors that have been proposed in the Bayesian instrumental variables literature. We show how to calculate the probability of any relevant restriction (e.g. the posterior probability that over-identifying restrictions hold) and discuss diagnostic checking using the posterior distribution of discrepancy vectors. We illustrate our methods in a returns-to-schooling application.

Predictors of temporary and permanent work disability in patients with inflammatory bowel disease: results of the swiss inflammatory bowel disease cohort study.

BACKGROUND: Inflammatory bowel disease can decrease the quality of life and induce work disability. We sought to (1) identify and quantify the predictors of disease-specific work disability in patients with inflammatory bowel disease and (2) assess the suitability of using cross-sectional data to predict future outcomes, using the Swiss Inflammatory Bowel Disease Cohort Study data. METHODS: A total of 1187 patients were enrolled and followed up for an average of 13 months. Predictors included patient and disease characteristics and drug utilization. Potential predictors were identified through an expert panel and published literature. We estimated adjusted effect estimates with 95% confidence intervals using logistic and zero-inflated Poisson regression. RESULTS: Overall, 699 (58.9%) experienced Crohn's disease and 488 (41.1%) had ulcerative colitis. Most important predictors for temporary work disability in patients with Crohn's disease included gender, disease duration, disease activity, C-reactive protein level, smoking, depressive symptoms, fistulas, extraintestinal manifestations, and the use of immunosuppressants/steroids. Temporary work disability in patients with ulcerative colitis was associated with age, disease duration, disease activity, and the use of steroids/antibiotics. In all patients, disease activity emerged as the only predictor of permanent work disability. Comparing data at enrollment versus follow-up yielded substantial differences regarding disability and predictors, with follow-up data showing greater predictor effects. CONCLUSIONS: We identified predictors of work disability in patients with Crohn's disease and ulcerative colitis. Our findings can help in forecasting these disease courses and guide the choice of appropriate measures to prevent adverse outcomes. Comparing cross-sectional and longitudinal data showed that the conduction of cohort studies is inevitable for the examination of disability.

Reversible behavioural autistic-like regression: A manifestation of a special (new?) epileptic syndrome in a 28-month-old child. A 2-year longitudinal study .

A 28-month-old boy was referred for acute onset of abnormal head movements. History revealed an insidious progressive regression in behaviour and communication over several months. Head and shoulder 'spasms' with alteration of consciousness and on one occasion ictal laughter were seen. The electroencephalograph (EEG) showed repeated bursts of brief generalized polyspikes and spike-wave during the 'spasms', followed by flattening, a special pattern which never recurred after treatment. Review of family videos showed a single 'minor' identical seizure 6 months previously. Magnetic resonance imaging was normal. Clonazepam brought immediate cessation of seizures, normalization of the EEG and a parallel spectacular improvement in communication, mood and language. Follow-up over the next 10 months showed a new regression unaccompained by recognized seizures, although numerous seizures were discovered during the videotaped neuropsychological examination, when stereotyped subtle brief paroxysmal changes in posture and behaviour could be studied in slow motion and compared with the 'prototypical' initial ones. The EEG showed predominant rare left-sided fronto-temporal discharges. Clonazepam was changed to carbamazepin with marked improvement in behaviour, language and cognition which has been sustained up to the last control at 51 months. Videotaped home observations allowed the documentation of striking qualitative and quantitative variations in social interaction and play of autistic type in relation to the epileptic activity. We conclude that this child has a special characteristic epileptic syndrome with subtle motor and vegetative symptomatology associated with an insidious catastrophic 'autistic-like' regression which could be overlooked. The methods used to document such fluctuating epileptic behavioural manifestations are discussed.

Model Switching and Model Averaging in Time- Varying Parameter Regression Models

This paper investigates the usefulness of switching Gaussian state space models as a tool for implementing dynamic model selecting (DMS) or averaging (DMA) in time-varying parameter regression models. DMS methods allow for model switching, where a different model can be chosen at each point in time. Thus, they allow for the explanatory variables in the time-varying parameter regression model to change over time. DMA will carry out model averaging in a time-varying manner. We compare our exact approach to DMA/DMS to a popular existing procedure which relies on the use of forgetting factor approximations. In an application, we use DMS to select different predictors in an in ation forecasting application. We also compare different ways of implementing DMA/DMS and investigate whether they lead to similar results.

Estimation of lean meat content in pig carcasses using X-ray Computed Tomography and PLS regression

Lean meat percentage (LMP) is an important carcass quality parameter. The aim of this work is to obtain a calibration equation for the Computed Tomography (CT) scans with the Partial Least Square Regression (PLS) technique in order to predict the LMP of the carcass and the different cuts and to study and compare two different methodologies of the selection of the variables (Variable Importance for Projection — VIP- and Stepwise) to be included in the prediction equation. The error of prediction with cross-validation (RMSEPCV) of the LMP obtained with PLS and selection based on VIP value was 0.82% and for stepwise selection it was 0.83%. The prediction of the LMP scanning only the ham had a RMSEPCV of 0.97% and if the ham and the loin were scanned the RMSEPCV was 0.90%. Results indicate that for CT data both VIP and stepwise selection are good methods. Moreover the scanning of only the ham allowed us to obtain a good prediction of the LMP of the whole carcass.

Quantile forecasts of inflation under model uncertainty

Bayesian model averaging (BMA) methods are regularly used to deal with model uncertainty in regression models. This paper shows how to introduce Bayesian model averaging methods in quantile regressions, and allow for different predictors to affect different quantiles of the dependent variable. I show that quantile regression BMA methods can help reduce uncertainty regarding outcomes of future inflation by providing superior predictive densities compared to mean regression models with and without BMA.

Innovation sources and productivity in Catalonian firms: a quantile regression analysis

This paper explores the effects of two main sources of innovation - intramural and external R&D— on the productivity level in a sample of 3,267 Catalonian firms. The data set used is based on the official innovation survey of Catalonia which was a part of the Spanish sample of CIS4, covering the years 2002-2004. We compare empirical results by applying usual OLS and quantile regression techniques both in manufacturing and services industries. In quantile regression, results suggest different patterns at both innovation sources as we move across conditional quantiles. The elasticity of intramural R&D activities on productivity decreased when we move up the high productivity levels both in manufacturing and services sectors, while the effects of external R&D rise in high-technology industries but are more ambiguous in low-technology and knowledge-intensive services. JEL codes: O300, C100, O140 Keywords: Innovation sources, R&D, Productivity, Quantile Regression

Is intra-operative blood flow predictive for early failure of radiocephalic arteriovenous fistula?

BACKGROUND: For over 50 years, radiocephalic wrist arteriovenous fistulae (RCAVF) have been the primary and best vascular access for haemodialysis. Nevertheless, early failure due to thrombosis or non-maturation is a major complication resulting in their abandonment. This prospective study was designed to investigate the predictive value of intra-operative blood flow on early failure of primary RCAVF before the first effective dialysis. METHODS: We enrolled patients undergoing creation of primary RCAVF for haemodialysis based on the pre-operative ultrasound vascular mapping discussed in a multidisciplinary approach. Intra-operative blood flow measurement was systematically performed once the anastomosis had been completed using a transit-time ultrasonic flowmeter. During the follow-up, blood flow was estimated by colour flow ultrasound at various intervals. Any events related to the RCAVF were recorded. RESULTS: Autogenous RCAVFs (n = 58) in 58 patients were constructed and followed up for an average of 30 days. Thrombosis and non-maturation occurred in eight (14%) and four (7%) patients, respectively. The intra-operative blood flow in functioning RCAVFs was significantly higher compared to non-functioning RCAVFs (230 vs 98 mL/min; P = 0.007), as well as 1 week (753 vs 228 mL/min; P = 0.0008) and 4 weeks (915 vs 245 mL/min, P < 0.0001) later. Blood flow volume measurements with a cut-off value of 120 mL/min had a sensitivity of 67%, specificity of 75% and positive predictive value of 91%. CONCLUSIONS: Blood flow <120 mL has a good predictive value for early failure in RCAVF. During the procedure, this cut-off value may be used to select appropriately which RCAVF should be investigated in the operation theatre in order to correct in real time any abnormality.

Modelling dependence in a ratemaking procedure with multivariate Poisson regression models

When actuaries face with the problem of pricing an insurance contract that contains different types of coverage, such as a motor insurance or homeowner's insurance policy, they usually assume that types of claim are independent. However, this assumption may not be realistic: several studies have shown that there is a positive correlation between types of claim. Here we introduce different regression models in order to relax the independence assumption, including zero-inflated models to account for excess of zeros and overdispersion. These models have been largely ignored to multivariate Poisson date, mainly because of their computational di±culties. Bayesian inference based on MCMC helps to solve this problem (and also lets us derive, for several quantities of interest, posterior summaries to account for uncertainty). Finally, these models are applied to an automobile insurance claims database with three different types of claims. We analyse the consequences for pure and loaded premiums when the independence assumption is relaxed by using different multivariate Poisson regression models and their zero-inflated versions.

Evaluating the influence of spatial uncertainty in locality points for species distributional modeling

1. Species distribution modelling is used increasingly in both applied and theoretical research to predict how species are distributed and to understand attributes of species' environmental requirements. In species distribution modelling, various statistical methods are used that combine species occurrence data with environmental spatial data layers to predict the suitability of any site for that species. While the number of data sharing initiatives involving species' occurrences in the scientific community has increased dramatically over the past few years, various data quality and methodological concerns related to using these data for species distribution modelling have not been addressed adequately. 2. We evaluated how uncertainty in georeferences and associated locational error in occurrences influence species distribution modelling using two treatments: (1) a control treatment where models were calibrated with original, accurate data and (2) an error treatment where data were first degraded spatially to simulate locational error. To incorporate error into the coordinates, we moved each coordinate with a random number drawn from the normal distribution with a mean of zero and a standard deviation of 5 km. We evaluated the influence of error on the performance of 10 commonly used distributional modelling techniques applied to 40 species in four distinct geographical regions. 3. Locational error in occurrences reduced model performance in three of these regions; relatively accurate predictions of species distributions were possible for most species, even with degraded occurrences. Two species distribution modelling techniques, boosted regression trees and maximum entropy, were the best performing models in the face of locational errors. The results obtained with boosted regression trees were only slightly degraded by errors in location, and the results obtained with the maximum entropy approach were not affected by such errors. 4. Synthesis and applications. To use the vast array of occurrence data that exists currently for research and management relating to the geographical ranges of species, modellers need to know the influence of locational error on model quality and whether some modelling techniques are particularly robust to error. We show that certain modelling techniques are particularly robust to a moderate level of locational error and that useful predictions of species distributions can be made even when occurrence data include some error.

MMP-9 as predictive factor for response and progression free survival in breast cancer patients treated with bevacizumab and pegylated liposomal doxorubicin. A multicenter, single-arm phase II trial (SAKK24/06). On behalf of the Swiss Group for Clinical Cancer Research (SAKK).

The benefit of bevacizumab (Bv) has been shown in different tumors including colorectal cancer, renal cancer, pulmonary non-small cell cancer and also breast cancer. However to date, there is no established test evaluating the angiogenic status of a patient and monitoring the effects of anti-angiogenic treatments. Tumor angiogenesis is the result of a balance between multiple pro- and anti¬angiogenic molecules. There is very little published clinical data exploring the impact of the anti-angiogenic therapy on the different angiogenesis-related molecules and the potential role of these molecules as prognostic or predictive factors.

Predicting smoking cessation and its relapse in HIV-infected patients: the Swiss HIV Cohort Study.

OBJECTIVES: The aim of the study was to assess whether prospective follow-up data within the Swiss HIV Cohort Study can be used to predict patients who stop smoking; or among smokers who stop, those who start smoking again. METHODS: We built prediction models first using clinical reasoning ('clinical models') and then by selecting from numerous candidate predictors using advanced statistical methods ('statistical models'). Our clinical models were based on literature that suggests that motivation drives smoking cessation, while dependence drives relapse in those attempting to stop. Our statistical models were based on automatic variable selection using additive logistic regression with component-wise gradient boosting. RESULTS: Of 4833 smokers, 26% stopped smoking, at least temporarily; because among those who stopped, 48% started smoking again. The predictive performance of our clinical and statistical models was modest. A basic clinical model for cessation, with patients classified into three motivational groups, was nearly as discriminatory as a constrained statistical model with just the most important predictors (the ratio of nonsmoking visits to total visits, alcohol or drug dependence, psychiatric comorbidities, recent hospitalization and age). A basic clinical model for relapse, based on the maximum number of cigarettes per day prior to stopping, was not as discriminatory as a constrained statistical model with just the ratio of nonsmoking visits to total visits. CONCLUSIONS: Predicting smoking cessation and relapse is difficult, so that simple models are nearly as discriminatory as complex ones. Patients with a history of attempting to stop and those known to have stopped recently are the best candidates for an intervention.