Robust parametric indirect estimates of the expected cost of a hospital stay with covariates and censored data.

We consider the problem of estimating the mean hospital cost of stays of a class of patients (e.g., a diagnosis-related group) as a function of patient characteristics. The statistical analysis is complicated by the asymmetry of the cost distribution, the possibility of censoring on the cost variable, and the occurrence of outliers. These problems have often been treated separately in the literature, and a method offering a joint solution to all of them is still missing. Indirect procedures have been proposed, combining an estimate of the duration distribution with an estimate of the conditional cost for a given duration. We propose a parametric version of this approach, allowing for asymmetry and censoring in the cost distribution and providing a mean cost estimator that is robust in the presence of extreme values. In addition, the new method takes covariate information into account.

An Italian to Catalan RBMT system reusing data from existing language pairs

This paper presents an Italian to CatalanRBMT system automatically built bycombining the linguistic data of theexisting pairs Spanish-Catalan andSpanish-Italian. A lightweight manualpostprocessing is carried out in order tofix inconsistencies in the automaticallyderived dictionaries and to add very frequentwords that are missing accordingto a corpus analysis. The system isevaluated on the KDE4 corpus and outperformsGoogle Translate by approximatelyten absolute points in terms ofboth TER and GTM.

Meta-analysis and imputation refines the association of 15q25 with smoking quantity.

Smoking is a leading global cause of disease and mortality. We established the Oxford-GlaxoSmithKline study (Ox-GSK) to perform a genome-wide meta-analysis of SNP association with smoking-related behavioral traits. Our final data set included 41,150 individuals drawn from 20 disease, population and control cohorts. Our analysis confirmed an effect on smoking quantity at a locus on 15q25 (P = 9.45 x 10(-19)) that includes CHRNA5, CHRNA3 and CHRNB4, three genes encoding neuronal nicotinic acetylcholine receptor subunits. We used data from the 1000 Genomes project to investigate the region using imputation, which allowed for analysis of virtually all common SNPs in the region and offered a fivefold increase in marker density over HapMap2 (ref. 2) as an imputation reference panel. Our fine-mapping approach identified a SNP showing the highest significance, rs55853698, located within the promoter region of CHRNA5. Conditional analysis also identified a secondary locus (rs6495308) in CHRNA3.

Responding to Young People's Health Risks in Primary Care: A Cluster Randomised Trial of Training Clinicians in Screening and Motivational Interviewing.

OBJECTIVE: To evaluate the effectiveness of a complex intervention implementing best practice guidelines recommending clinicians screen and counsel young people across multiple psychosocial risk factors, on clinicians' detection of health risks and patients' risk taking behaviour, compared to a didactic seminar on young people's health. DESIGN: Pragmatic cluster randomised trial where volunteer general practices were stratified by postcode advantage or disadvantage score and billing type (private, free national health, community health centre), then randomised into either intervention or comparison arms using a computer generated random sequence. Three months post-intervention, patients were recruited from all practices post-consultation for a Computer Assisted Telephone Interview and followed up three and 12 months later. Researchers recruiting, consenting and interviewing patients and patients themselves were masked to allocation status; clinicians were not. SETTING: General practices in metropolitan and rural Victoria, Australia. PARTICIPANTS: General practices with at least one interested clinician (general practitioner or nurse) and their 14-24 year old patients. INTERVENTION: This complex intervention was designed using evidence based practice in learning and change in clinician behaviour and general practice systems, and included best practice approaches to motivating change in adolescent risk taking behaviours. The intervention involved training clinicians (nine hours) in health risk screening, use of a screening tool and motivational interviewing; training all practice staff (receptionists and clinicians) in engaging youth; provision of feedback to clinicians of patients' risk data; and two practice visits to support new screening and referral resources. Comparison clinicians received one didactic educational seminar (three hours) on engaging youth and health risk screening. OUTCOME MEASURES: Primary outcomes were patient report of (1) clinician detection of at least one of six health risk behaviours (tobacco, alcohol and illicit drug use, risks for sexually transmitted infection, STI, unplanned pregnancy, and road risks); and (2) change in one or more of the six health risk behaviours, at three months or at 12 months. Secondary outcomes were likelihood of future visits, trust in the clinician after exit interview, clinician detection of emotional distress and fear and abuse in relationships, and emotional distress at three and 12 months. Patient acceptability of the screening tool was also described for the intervention arm. Analyses were adjusted for practice location and billing type, patients' sex, age, and recruitment method, and past health risks, where appropriate. An intention to treat analysis approach was used, which included multilevel multiple imputation for missing outcome data. RESULTS: 42 practices were randomly allocated to intervention or comparison arms. Two intervention practices withdrew post allocation, prior to training, leaving 19 intervention (53 clinicians, 377 patients) and 21 comparison (79 clinicians, 524 patients) practices. 69% of patients in both intervention (260) and comparison (360) arms completed the 12 month follow-up. Intervention clinicians discussed more health risks per patient (59.7%) than comparison clinicians (52.7%) and thus were more likely to detect a higher proportion of young people with at least one of the six health risk behaviours (38.4% vs 26.7%, risk difference [RD] 11.6%, Confidence Interval [CI] 2.93% to 20.3%; adjusted odds ratio [OR] 1.7, CI 1.1 to 2.5). Patients reported less illicit drug use (RD -6.0, CI -11 to -1.2; OR 0·52, CI 0·28 to 0·96), and less risk for STI (RD -5.4, CI -11 to 0.2; OR 0·66, CI 0·46 to 0·96) at three months in the intervention relative to the comparison arm, and for unplanned pregnancy at 12 months (RD -4.4; CI -8.7 to -0.1; OR 0·40, CI 0·20 to 0·80). No differences were detected between arms on other health risks. There were no differences on secondary outcomes, apart from a greater detection of abuse (OR 13.8, CI 1.71 to 111). There were no reports of harmful events and intervention arm youth had high acceptance of the screening tool. CONCLUSIONS: A complex intervention, compared to a simple educational seminar for practices, improved detection of health risk behaviours in young people. Impact on health outcomes was inconclusive. Technology enabling more efficient, systematic health-risk screening may allow providers to target counselling toward higher risk individuals. Further trials require more power to confirm health benefits. TRIAL REGISTRATION: ISRCTN.com ISRCTN16059206.

Assessing the use of very high resolution data to predict species distribution in a mountain environment

Nowadays, Species Distribution Models (SDMs) are a widely used tool. Using different statistical approaches these models reconstruct the realized niche of a species using presence data and a set of variables, often topoclimatic. There utilization range is quite large from understanding single species requirements, to the creation of nature reserve based on species hotspots, or modeling of climate change impact, etc... Most of the time these models are using variables at a resolution of 50km x 50km or 1 km x 1 km. However in some cases these models are used with resolutions below the kilometer scale and thus called high resolution models (100 m x 100 m or 25 m x 25 m). Quite recently a new kind of data has emerged enabling precision up to lm x lm and thus allowing very high resolution modeling. However these new variables are very costly and need an important amount of time to be processed. This is especially the case when these variables are used in complex calculation like models projections over large areas. Moreover the importance of very high resolution data in SDMs has not been assessed yet and is not well understood. Some basic knowledge on what drive species presence-absences is still missing. Indeed, it is not clear whether in mountain areas like the Alps coarse topoclimatic gradients are driving species distributions or if fine scale temperature or topography are more important or if their importance can be neglected when balance to competition or stochasticity. In this thesis I investigated the importance of very high resolution data (2-5m) in species distribution models using either very high resolution topographic, climatic or edaphic variables over a 2000m elevation gradient in the Western Swiss Alps. I also investigated more local responses of these variables for a subset of species living in this area at two precise elvation belts. During this thesis I showed that high resolution data necessitates very good datasets (species and variables for the models) to produce satisfactory results. Indeed, in mountain areas, temperature is the most important factor driving species distribution and needs to be modeled at very fine resolution instead of being interpolated over large surface to produce satisfactory results. Despite the instinctive idea that topographic should be very important at high resolution, results are mitigated. However looking at the importance of variables over a large gradient buffers the importance of the variables. Indeed topographic factors have been shown to be highly important at the subalpine level but their importance decrease at lower elevations. Wether at the mountane level edaphic and land use factors are more important high resolution topographic data is more imporatant at the subalpine level. Finally the biggest improvement in the models happens when edaphic variables are added. Indeed, adding soil variables is of high importance and variables like pH are overpassing the usual topographic variables in SDMs in term of importance in the models. To conclude high resolution is very important in modeling but necessitate very good datasets. Only increasing the resolution of the usual topoclimatic predictors is not sufficient and the use of edaphic predictors has been highlighted as fundamental to produce significantly better models. This is of primary importance, especially if these models are used to reconstruct communities or as basis for biodiversity assessments. -- Ces dernières années, l'utilisation des modèles de distribution d'espèces (SDMs) a continuellement augmenté. Ces modèles utilisent différents outils statistiques afin de reconstruire la niche réalisée d'une espèce à l'aide de variables, notamment climatiques ou topographiques, et de données de présence récoltées sur le terrain. Leur utilisation couvre de nombreux domaines allant de l'étude de l'écologie d'une espèce à la reconstruction de communautés ou à l'impact du réchauffement climatique. La plupart du temps, ces modèles utilisent des occur-rences issues des bases de données mondiales à une résolution plutôt large (1 km ou même 50 km). Certaines bases de données permettent cependant de travailler à haute résolution, par conséquent de descendre en dessous de l'échelle du kilomètre et de travailler avec des résolutions de 100 m x 100 m ou de 25 m x 25 m. Récemment, une nouvelle génération de données à très haute résolution est apparue et permet de travailler à l'échelle du mètre. Les variables qui peuvent être générées sur la base de ces nouvelles données sont cependant très coûteuses et nécessitent un temps conséquent quant à leur traitement. En effet, tout calcul statistique complexe, comme des projections de distribution d'espèces sur de larges surfaces, demande des calculateurs puissants et beaucoup de temps. De plus, les facteurs régissant la distribution des espèces à fine échelle sont encore mal connus et l'importance de variables à haute résolution comme la microtopographie ou la température dans les modèles n'est pas certaine. D'autres facteurs comme la compétition ou la stochasticité naturelle pourraient avoir une influence toute aussi forte. C'est dans ce contexte que se situe mon travail de thèse. J'ai cherché à comprendre l'importance de la haute résolution dans les modèles de distribution d'espèces, que ce soit pour la température, la microtopographie ou les variables édaphiques le long d'un important gradient d'altitude dans les Préalpes vaudoises. J'ai également cherché à comprendre l'impact local de certaines variables potentiellement négligées en raison d'effets confondants le long du gradient altitudinal. Durant cette thèse, j'ai pu monter que les variables à haute résolution, qu'elles soient liées à la température ou à la microtopographie, ne permettent qu'une amélioration substantielle des modèles. Afin de distinguer une amélioration conséquente, il est nécessaire de travailler avec des jeux de données plus importants, tant au niveau des espèces que des variables utilisées. Par exemple, les couches climatiques habituellement interpolées doivent être remplacées par des couches de température modélisées à haute résolution sur la base de données de terrain. Le fait de travailler le long d'un gradient de température de 2000m rend naturellement la température très importante au niveau des modèles. L'importance de la microtopographie est négligeable par rapport à la topographie à une résolution de 25m. Cependant, lorsque l'on regarde à une échelle plus locale, la haute résolution est une variable extrêmement importante dans le milieu subalpin. À l'étage montagnard par contre, les variables liées aux sols et à l'utilisation du sol sont très importantes. Finalement, les modèles de distribution d'espèces ont été particulièrement améliorés par l'addition de variables édaphiques, principalement le pH, dont l'importance supplante ou égale les variables topographique lors de leur ajout aux modèles de distribution d'espèces habituels.

Statistical analysis of survey-based event history data with application to modeling of unemployment duration

Longitudinal surveys are increasingly used to collect event history data on person-specific processes such as transitions between labour market states. Surveybased event history data pose a number of challenges for statistical analysis. These challenges include survey errors due to sampling, non-response, attrition and measurement. This study deals with non-response, attrition and measurement errors in event history data and the bias caused by them in event history analysis. The study also discusses some choices faced by a researcher using longitudinal survey data for event history analysis and demonstrates their effects. These choices include, whether a design-based or a model-based approach is taken, which subset of data to use and, if a design-based approach is taken, which weights to use. The study takes advantage of the possibility to use combined longitudinal survey register data. The Finnish subset of European Community Household Panel (FI ECHP) survey for waves 1–5 were linked at person-level with longitudinal register data. Unemployment spells were used as study variables of interest. Lastly, a simulation study was conducted in order to assess the statistical properties of the Inverse Probability of Censoring Weighting (IPCW) method in a survey data context. The study shows how combined longitudinal survey register data can be used to analyse and compare the non-response and attrition processes, test the missingness mechanism type and estimate the size of bias due to non-response and attrition. In our empirical analysis, initial non-response turned out to be a more important source of bias than attrition. Reported unemployment spells were subject to seam effects, omissions, and, to a lesser extent, overreporting. The use of proxy interviews tended to cause spell omissions. An often-ignored phenomenon classification error in reported spell outcomes, was also found in the data. Neither the Missing At Random (MAR) assumption about non-response and attrition mechanisms, nor the classical assumptions about measurement errors, turned out to be valid. Both measurement errors in spell durations and spell outcomes were found to cause bias in estimates from event history models. Low measurement accuracy affected the estimates of baseline hazard most. The design-based estimates based on data from respondents to all waves of interest and weighted by the last wave weights displayed the largest bias. Using all the available data, including the spells by attriters until the time of attrition, helped to reduce attrition bias. Lastly, the simulation study showed that the IPCW correction to design weights reduces bias due to dependent censoring in design-based Kaplan-Meier and Cox proportional hazard model estimators. The study discusses implications of the results for survey organisations collecting event history data, researchers using surveys for event history analysis, and researchers who develop methods to correct for non-sampling biases in event history data.

Attribute reduction and missing value imputing with ANN: prediction of learning disabilities

Learning disability (LD) is a neurological condition that affects a child’s brain and impairs his ability to carry out one or many specific tasks. LD affects about 10% of children enrolled in schools. There is no cure for learning disabilities and they are lifelong. The problems of children with specific learning disabilities have been a cause of concern to parents and teachers for some time. Just as there are many different types of LDs, there are a variety of tests that may be done to pinpoint the problem The information gained from an evaluation is crucial for finding out how the parents and the school authorities can provide the best possible learning environment for child. This paper proposes a new approach in artificial neural network (ANN) for identifying LD in children at early stages so as to solve the problems faced by them and to get the benefits to the students, their parents and school authorities. In this study, we propose a closest fit algorithm data preprocessing with ANN classification to handle missing attribute values. This algorithm imputes the missing values in the preprocessing stage. Ignoring of missing attribute values is a common trend in all classifying algorithms. But, in this paper, we use an algorithm in a systematic approach for classification, which gives a satisfactory result in the prediction of LD. It acts as a tool for predicting the LD accurately, and good information of the child is made available to the concerned

Modelling structural zeros in compositional data

This analysis was stimulated by the real data analysis problem of household expenditure data. The full dataset contains expenditure data for a sample of 1224 households. The expenditure is broken down at 2 hierarchical levels: 9 major levels (e.g. housing, food, utilities etc.) and 92 minor levels. There are also 5 factors and 5 covariates at the household level. Not surprisingly, there are a small number of zeros at the major level, but many zeros at the minor level. The question is how best to model the zeros. Clearly, models that try to add a small amount to the zero terms are not appropriate in general as at least some of the zeros are clearly structural, e.g. alcohol/tobacco for households that are teetotal. The key question then is how to build suitable conditional models. For example, is the sub-composition of spending excluding alcohol/tobacco similar for teetotal and non-teetotal households? In other words, we are looking for sub-compositional independence. Also, what determines whether a household is teetotal? Can we assume that it is independent of the composition? In general, whether teetotal will clearly depend on the household level variables, so we need to be able to model this dependence. The other tricky question is that with zeros on more than one component, we need to be able to model dependence and independence of zeros on the different components. Lastly, while some zeros are structural, others may not be, for example, for expenditure on durables, it may be chance as to whether a particular household spends money on durables within the sample period. This would clearly be distinguishable if we had longitudinal data, but may still be distinguishable by looking at the distribution, on the assumption that random zeros will usually be for situations where any non-zero expenditure is not small. While this analysis is based on around economic data, the ideas carry over to many other situations, including geological data, where minerals may be missing for structural reasons (similar to alcohol), or missing because they occur only in random regions which may be missed in a sample (similar to the durables)

Compositional Data Analysis with R

R from http://www.r-project.org/ is ‘GNU S’ – a language and environment for statistical computing and graphics. The environment in which many classical and modern statistical techniques have been implemented, but many are supplied as packages. There are 8 standard packages and many more are available through the cran family of Internet sites http://cran.r-project.org . We started to develop a library of functions in R to support the analysis of mixtures and our goal is a MixeR package for compositional data analysis that provides support for operations on compositions: perturbation and power multiplication, subcomposition with or without residuals, centering of the data, computing Aitchison’s, Euclidean, Bhattacharyya distances, compositional Kullback-Leibler divergence etc. graphical presentation of compositions in ternary diagrams and tetrahedrons with additional features: barycenter, geometric mean of the data set, the percentiles lines, marking and coloring of subsets of the data set, theirs geometric means, notation of individual data in the set . . . dealing with zeros and missing values in compositional data sets with R procedures for simple and multiplicative replacement strategy, the time series analysis of compositional data. We’ll present the current status of MixeR development and illustrate its use on selected data sets

Preparation of consistent soil data sets for modelling purposes: Secondary SOTER data for four case study areas

The common GIS-based approach to regional analyses of soil organic carbon (SOC) stocks and changes is to define geographic layers for which unique sets of driving variables are derived, which include land use, climate, and soils. These GIS layers, with their associated attribute data, can then be fed into a range of empirical and dynamic models. Common methodologies for collating and formatting regional data sets on land use, climate, and soils were adopted for the project Assessment of Soil Organic Carbon Stocks and Changes at National Scale (GEFSOC). This permitted the development of a uniform protocol for handling the various input for the dynamic GEFSOC Modelling System. Consistent soil data sets for Amazon-Brazil, the Indo-Gangetic Plains (IGP) of India, Jordan and Kenya, the case study areas considered in the GEFSOC project, were prepared using methodologies developed for the World Soils and Terrain Database (SOTER). The approach involved three main stages: (1) compiling new soil geographic and attribute data in SOTER format; (2) using expert estimates and common sense to fill selected gaps in the measured or primary data; (3) using a scheme of taxonomy-based pedotransfer rules and expert-rules to derive soil parameter estimates for similar soil units with missing soil analytical data. The most appropriate approach varied from country to country, depending largely on the overall accessibility and quality of the primary soil data available in the case study areas. The secondary SOTER data sets discussed here are appropriate for a wide range of environmental applications at national scale. These include agro-ecological zoning, land evaluation, modelling of soil C stocks and changes, and studies of soil vulnerability to pollution. Estimates of national-scale stocks of SOC, calculated using SOTER methods, are presented as a first example of database application. Independent estimates of SOC stocks are needed to evaluate the outcome of the GEFSOC Modelling System for current conditions of land use and climate. (C) 2007 Elsevier B.V. All rights reserved.

Automated pre-processing strategies for species occurrence data used in biodiversity modelling

To construct Biodiversity richness maps from Environmental Niche Models (ENMs) of thousands of species is time consuming. A separate species occurrence data pre-processing phase enables the experimenter to control test AUC score variance due to species dataset size. Besides, removing duplicate occurrences and points with missing environmental data, we discuss the need for coordinate precision, wide dispersion, temporal and synonymity filters. After species data filtering, the final task of a pre-processing phase should be the automatic generation of species occurrence datasets which can then be directly ’plugged-in’ to the ENM. A software application capable of carrying out all these tasks will be a valuable time-saver particularly for large scale biodiversity studies.

Mixture models for capture-recapture count data

The contribution investigates the problem of estimating the size of a population, also known as the missing cases problem. Suppose a registration system is targeting to identify all cases having a certain characteristic such as a specific disease (cancer, heart disease, ...), disease related condition (HIV, heroin use, ...) or a specific behavior (driving a car without license). Every case in such a registration system has a certain notification history in that it might have been identified several times (at least once) which can be understood as a particular capture-recapture situation. Typically, cases are left out which have never been listed at any occasion, and it is this frequency one wants to estimate. In this paper modelling is concentrating on the counting distribution, e.g. the distribution of the variable that counts how often a given case has been identified by the registration system. Besides very simple models like the binomial or Poisson distribution, finite (nonparametric) mixtures of these are considered providing rather flexible modelling tools. Estimation is done using maximum likelihood by means of the EM algorithm. A case study on heroin users in Bangkok in the year 2001 is completing the contribution.

Extending Zelterman’s approach for robust estimation of population size to zero-truncated clustered data

Estimation of population size with missing zero-class is an important problem that is encountered in epidemiological assessment studies. Fitting a Poisson model to the observed data by the method of maximum likelihood and estimation of the population size based on this fit is an approach that has been widely used for this purpose. In practice, however, the Poisson assumption is seldom satisfied. Zelterman (1988) has proposed a robust estimator for unclustered data that works well in a wide class of distributions applicable for count data. In the work presented here, we extend this estimator to clustered data. The estimator requires fitting a zero-truncated homogeneous Poisson model by maximum likelihood and thereby using a Horvitz-Thompson estimator of population size. This was found to work well, when the data follow the hypothesized homogeneous Poisson model. However, when the true distribution deviates from the hypothesized model, the population size was found to be underestimated. In the search of a more robust estimator, we focused on three models that use all clusters with exactly one case, those clusters with exactly two cases and those with exactly three cases to estimate the probability of the zero-class and thereby use data collected on all the clusters in the Horvitz-Thompson estimator of population size. Loss in efficiency associated with gain in robustness was examined based on a simulation study. As a trade-off between gain in robustness and loss in efficiency, the model that uses data collected on clusters with at most three cases to estimate the probability of the zero-class was found to be preferred in general. In applications, we recommend obtaining estimates from all three models and making a choice considering the estimates from the three models, robustness and the loss in efficiency. (© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)

PDM 2006: Proceedings of the International Workshop on Data Mining

Recently major processor manufacturers have announced a dramatic shift in their paradigm to increase computing power over the coming years. Instead of focusing on faster clock speeds and more powerful single core CPUs, the trend clearly goes towards multi core systems. This will also result in a paradigm shift for the development of algorithms for computationally expensive tasks, such as data mining applications. Obviously, work on parallel algorithms is not new per se but concentrated efforts in the many application domains are still missing. Multi-core systems, but also clusters of workstations and even large-scale distributed computing infrastructures provide new opportunities and pose new challenges for the design of parallel and distributed algorithms. Since data mining and machine learning systems rely on high performance computing systems, research on the corresponding algorithms must be on the forefront of parallel algorithm research in order to keep pushing data mining and machine learning applications to be more powerful and, especially for the former, interactive. To bring together researchers and practitioners working in this exciting field, a workshop on parallel data mining was organized as part of PKDD/ECML 2006 (Berlin, Germany). The six contributions selected for the program describe various aspects of data mining and machine learning approaches featuring low to high degrees of parallelism: The first contribution focuses the classic problem of distributed association rule mining and focuses on communication efficiency to improve the state of the art. After this a parallelization technique for speeding up decision tree construction by means of thread-level parallelism for shared memory systems is presented. The next paper discusses the design of a parallel approach for dis- tributed memory systems of the frequent subgraphs mining problem. This approach is based on a hierarchical communication topology to solve issues related to multi-domain computational envi- ronments. The forth paper describes the combined use and the customization of software packages to facilitate a top down parallelism in the tuning of Support Vector Machines (SVM) and the next contribution presents an interesting idea concerning parallel training of Conditional Random Fields (CRFs) and motivates their use in labeling sequential data. The last contribution finally focuses on very efficient feature selection. It describes a parallel algorithm for feature selection from random subsets. Selecting the papers included in this volume would not have been possible without the help of an international Program Committee that has provided detailed reviews for each paper. We would like to also thank Matthew Otey who helped with publicity for the workshop.

An Earth Observation Land Data Assimilation System (EO-LDAS)

Current methods for estimating vegetation parameters are generally sub-optimal in the way they exploit information and do not generally consider uncertainties. We look forward to a future where operational dataassimilation schemes improve estimates by tracking land surface processes and exploiting multiple types of observations. Dataassimilation schemes seek to combine observations and models in a statistically optimal way taking into account uncertainty in both, but have not yet been much exploited in this area. The EO-LDAS scheme and prototype, developed under ESA funding, is designed to exploit the anticipated wealth of data that will be available under GMES missions, such as the Sentinel family of satellites, to provide improved mapping of land surface biophysical parameters. This paper describes the EO-LDAS implementation, and explores some of its core functionality. EO-LDAS is a weak constraint variational dataassimilationsystem. The prototype provides a mechanism for constraint based on a prior estimate of the state vector, a linear dynamic model, and EarthObservationdata (top-of-canopy reflectance here). The observation operator is a non-linear optical radiative transfer model for a vegetation canopy with a soil lower boundary, operating over the range 400 to 2500 nm. Adjoint codes for all model and operator components are provided in the prototype by automatic differentiation of the computer codes. In this paper, EO-LDAS is applied to the problem of daily estimation of six of the parameters controlling the radiative transfer operator over the course of a year (> 2000 state vector elements). Zero and first order process model constraints are implemented and explored as the dynamic model. The assimilation estimates all state vector elements simultaneously. This is performed in the context of a typical Sentinel-2 MSI operating scenario, using synthetic MSI observations simulated with the observation operator, with uncertainties typical of those achieved by optical sensors supposed for the data. The experiments consider a baseline state vector estimation case where dynamic constraints are applied, and assess the impact of dynamic constraints on the a posteriori uncertainties. The results demonstrate that reductions in uncertainty by a factor of up to two might be obtained by applying the sorts of dynamic constraints used here. The hyperparameter (dynamic model uncertainty) required to control the assimilation are estimated by a cross-validation exercise. The result of the assimilation is seen to be robust to missing observations with quite large data gaps.