Assessing the Relationship between the Baseline Value of a Continuous Variable and Subsequent Change over Time

Analyzing the relationship between the baseline value and subsequent change of a continuous variable is a frequent matter of inquiry in cohort studies. These analyses are surprisingly complex, particularly if only two waves of data are available. It is unclear for non-biostatisticians where the complexity of this analysis lies and which statistical method is adequate.With the help of simulated longitudinal data of body mass index in children,we review statistical methods for the analysis of the association between the baseline value and subsequent change, assuming linear growth with time. Key issues in such analyses are mathematical coupling, measurement error, variability of change between individuals, and regression to the mean. Ideally, it is better to rely on multiple repeated measurements at different times and a linear random effects model is a standard approach if more than two waves of data are available. If only two waves of data are available, our simulations show that Blomqvist's method - which consists in adjusting for measurement error variance the estimated regression coefficient of observed change on baseline value - provides accurate estimates. The adequacy of the methods to assess the relationship between the baseline value and subsequent change depends on the number of data waves, the availability of information on measurement error, and the variability of change between individuals.

Correcting Measurement Error Bias in Interaction Models with Small Samples

Several methods have been suggested to estimate non-linear models with interaction terms in the presence of measurement error. Structural equation models eliminate measurement error bias, but require large samples. Ordinary least squares regression on summated scales, regression on factor scores and partial least squares are appropriate for small samples but do not correct measurement error bias. Two stage least squares regression does correct measurement error bias but the results strongly depend on the instrumental variable choice. This article discusses the old disattenuated regression method as an alternative for correcting measurement error in small samples. The method is extended to the case of interaction terms and is illustrated on a model that examines the interaction effect of innovation and style of use of budgets on business performance. Alternative reliability estimates that can be used to disattenuate the estimates are discussed. A comparison is made with the alternative methods. Methods that do not correct for measurement error bias perform very similarly and considerably worse than disattenuated regression

Kinematics and dynamic complexity of postural transitions in frail elderly subjects.

The aim of this study was to propose a methodology allowing a detailed characterization of body sit-to-stand/stand-to-sit postural transition. Parameters characterizing the kinematics of the trunk movement during sit-to-stand (Si-St) postural transition were calculated using one initial sensor system fixed on the trunk and a data logger. Dynamic complexity of these postural transitions was estimated by fractal dimension of acceleration-angular velocity plot. We concluded that this method provides a simple and accurate tool for monitoring frail elderly and to objectively evaluate the efficacy of a rehabilitation program.

Effective diagnosis of Alzheimer's disease by means of large margin-based methodology.

BACKGROUND Functional brain images such as Single-Photon Emission Computed Tomography (SPECT) and Positron Emission Tomography (PET) have been widely used to guide the clinicians in the Alzheimer's Disease (AD) diagnosis. However, the subjectivity involved in their evaluation has favoured the development of Computer Aided Diagnosis (CAD) Systems. METHODS It is proposed a novel combination of feature extraction techniques to improve the diagnosis of AD. Firstly, Regions of Interest (ROIs) are selected by means of a t-test carried out on 3D Normalised Mean Square Error (NMSE) features restricted to be located within a predefined brain activation mask. In order to address the small sample-size problem, the dimension of the feature space was further reduced by: Large Margin Nearest Neighbours using a rectangular matrix (LMNN-RECT), Principal Component Analysis (PCA) or Partial Least Squares (PLS) (the two latter also analysed with a LMNN transformation). Regarding the classifiers, kernel Support Vector Machines (SVMs) and LMNN using Euclidean, Mahalanobis and Energy-based metrics were compared. RESULTS Several experiments were conducted in order to evaluate the proposed LMNN-based feature extraction algorithms and its benefits as: i) linear transformation of the PLS or PCA reduced data, ii) feature reduction technique, and iii) classifier (with Euclidean, Mahalanobis or Energy-based methodology). The system was evaluated by means of k-fold cross-validation yielding accuracy, sensitivity and specificity values of 92.78%, 91.07% and 95.12% (for SPECT) and 90.67%, 88% and 93.33% (for PET), respectively, when a NMSE-PLS-LMNN feature extraction method was used in combination with a SVM classifier, thus outperforming recently reported baseline methods. CONCLUSIONS All the proposed methods turned out to be a valid solution for the presented problem. One of the advances is the robustness of the LMNN algorithm that not only provides higher separation rate between the classes but it also makes (in combination with NMSE and PLS) this rate variation more stable. In addition, their generalization ability is another advance since several experiments were performed on two image modalities (SPECT and PET).

Identification of pharmaceutical tablets by Raman spectroscopy and chemometrics

Raman spectroscopy has become an attractive tool for the analysis of pharmaceutical solid dosage forms. In the present study it is used to ensure the identity of tablets. The two main applications of this method are release of final products in quality control and detection of counterfeits. Twenty-five product families of tablets have been included in the spectral library and a non-linear classification method, the Support Vector Machines (SVMs), has been employed. Two calibrations have been developed in cascade: the first one identifies the product family while the second one specifies the formulation. A product family comprises different formulations that have the same active pharmaceutical ingredient (API) but in a different amount. Once the tablets have been classified by the SVM model, API peaks detection and correlation are applied in order to have a specific method for the identification and allow in the future to discriminate counterfeits from genuine products. This calibration strategy enables the identification of 25 product families without error and in the absence of prior information about the sample. Raman spectroscopy coupled with chemometrics is therefore a fast and accurate tool for the identification of pharmaceutical tablets.

3-d residual eddy current field characterisation: applied to diffusion weighted magnetic resonance imaging.

Clinical use of the Stejskal-Tanner diffusion weighted images is hampered by the geometric distortions that result from the large residual 3-D eddy current field induced. In this work, we aimed to predict, using linear response theory, the residual 3-D eddy current field required for geometric distortion correction based on phantom eddy current field measurements. The predicted 3-D eddy current field induced by the diffusion-weighting gradients was able to reduce the root mean square error of the residual eddy current field to ~1 Hz. The model's performance was tested on diffusion weighted images of four normal volunteers, following distortion correction, the quality of the Stejskal-Tanner diffusion-weighted images was found to have comparable quality to image registration based corrections (FSL) at low b-values. Unlike registration techniques the correction was not hindered by low SNR at high b-values, and results in improved image quality relative to FSL. Characterization of the 3-D eddy current field with linear response theory enables the prediction of the 3-D eddy current field required to correct eddy current induced geometric distortions for a wide range of clinical and high b-value protocols.

C-H···O H-bonded complexes: how does basis set superposition error change their potential-energy surfaces?

Geometries, vibrational frequencies, and interaction energies of the CNH⋯O3 and HCCH⋯O3 complexes are calculated in a counterpoise-corrected (CP-corrected) potential-energy surface (PES) that corrects for the basis set superposition error (BSSE). Ab initio calculations are performed at the Hartree-Fock (HF) and second-order Møller-Plesset (MP2) levels, using the 6-31G(d,p) and D95++(d,p) basis sets. Interaction energies are presented including corrections for zero-point vibrational energy (ZPVE) and thermal correction to enthalpy at 298 K. The CP-corrected and conventional PES are compared; the unconnected PES obtained using the larger basis set including diffuse functions exhibits a double well shape, whereas use of the 6-31G(d,p) basis set leads to a flat single-well profile. The CP-corrected PES has always a multiple-well shape. In particular, it is shown that the CP-corrected PES using the smaller basis set is qualitatively analogous to that obtained with the larger basis sets, so the CP method becomes useful to correctly describe large systems, where the use of small basis sets may be necessary

Dose reconstruction in the context of tomotherapy

In vivo dosimetry is a way to verify the radiation dose delivered to the patient in measuring the dose generally during the first fraction of the treatment. It is the only dose delivery control based on a measurement performed during the treatment. In today's radiotherapy practice, the dose delivered to the patient is planned using 3D dose calculation algorithms and volumetric images representing the patient. Due to the high accuracy and precision necessary in radiation treatments, national and international organisations like ICRU and AAPM recommend the use of in vivo dosimetry. It is also mandatory in some countries like France. Various in vivo dosimetry methods have been developed during the past years. These methods are point-, line-, plane- or 3D dose controls. A 3D in vivo dosimetry provides the most information about the dose delivered to the patient, with respect to ID and 2D methods. However, to our knowledge, it is generally not routinely applied to patient treatments yet. The aim of this PhD thesis was to determine whether it is possible to reconstruct the 3D delivered dose using transmitted beam measurements in the context of narrow beams. An iterative dose reconstruction method has been described and implemented. The iterative algorithm includes a simple 3D dose calculation algorithm based on the convolution/superposition principle. The methodology was applied to narrow beams produced by a conventional 6 MV linac. The transmitted dose was measured using an array of ion chambers, as to simulate the linear nature of a tomotherapy detector. We showed that the iterative algorithm converges quickly and reconstructs the dose within a good agreement (at least 3% / 3 mm locally), which is inside the 5% recommended by the ICRU. Moreover it was demonstrated on phantom measurements that the proposed method allows us detecting some set-up errors and interfraction geometry modifications. We also have discussed the limitations of the 3D dose reconstruction for dose delivery error detection. Afterwards, stability tests of the tomotherapy MVCT built-in onboard detector was performed in order to evaluate if such a detector is suitable for 3D in-vivo dosimetry. The detector showed stability on short and long terms comparable to other imaging devices as the EPIDs, also used for in vivo dosimetry. Subsequently, a methodology for the dose reconstruction using the tomotherapy MVCT detector is proposed in the context of static irradiations. This manuscript is composed of two articles and a script providing further information related to this work. In the latter, the first chapter introduces the state-of-the-art of in vivo dosimetry and adaptive radiotherapy, and explains why we are interested in performing 3D dose reconstructions. In chapter 2 a dose calculation algorithm implemented for this work is reviewed with a detailed description of the physical parameters needed for calculating 3D absorbed dose distributions. The tomotherapy MVCT detector used for transit measurements and its characteristics are described in chapter 3. Chapter 4 contains a first article entitled '3D dose reconstruction for narrow beams using ion chamber array measurements', which describes the dose reconstruction method and presents tests of the methodology on phantoms irradiated with 6 MV narrow photon beams. Chapter 5 contains a second article 'Stability of the Helical TomoTherapy HiArt II detector for treatment beam irradiations. A dose reconstruction process specific to the use of the tomotherapy MVCT detector is presented in chapter 6. A discussion and perspectives of the PhD thesis are presented in chapter 7, followed by a conclusion in chapter 8. The tomotherapy treatment device is described in appendix 1 and an overview of 3D conformai- and intensity modulated radiotherapy is presented in appendix 2. - La dosimétrie in vivo est une technique utilisée pour vérifier la dose délivrée au patient en faisant une mesure, généralement pendant la première séance du traitement. Il s'agit de la seule technique de contrôle de la dose délivrée basée sur une mesure réalisée durant l'irradiation du patient. La dose au patient est calculée au moyen d'algorithmes 3D utilisant des images volumétriques du patient. En raison de la haute précision nécessaire lors des traitements de radiothérapie, des organismes nationaux et internationaux tels que l'ICRU et l'AAPM recommandent l'utilisation de la dosimétrie in vivo, qui est devenue obligatoire dans certains pays dont la France. Diverses méthodes de dosimétrie in vivo existent. Elles peuvent être classées en dosimétrie ponctuelle, planaire ou tridimensionnelle. La dosimétrie 3D est celle qui fournit le plus d'information sur la dose délivrée. Cependant, à notre connaissance, elle n'est généralement pas appliquée dans la routine clinique. Le but de cette recherche était de déterminer s'il est possible de reconstruire la dose 3D délivrée en se basant sur des mesures de la dose transmise, dans le contexte des faisceaux étroits. Une méthode itérative de reconstruction de la dose a été décrite et implémentée. L'algorithme itératif contient un algorithme simple basé sur le principe de convolution/superposition pour le calcul de la dose. La dose transmise a été mesurée à l'aide d'une série de chambres à ionisations alignées afin de simuler la nature linéaire du détecteur de la tomothérapie. Nous avons montré que l'algorithme itératif converge rapidement et qu'il permet de reconstruire la dose délivrée avec une bonne précision (au moins 3 % localement / 3 mm). De plus, nous avons démontré que cette méthode permet de détecter certaines erreurs de positionnement du patient, ainsi que des modifications géométriques qui peuvent subvenir entre les séances de traitement. Nous avons discuté les limites de cette méthode pour la détection de certaines erreurs d'irradiation. Par la suite, des tests de stabilité du détecteur MVCT intégré à la tomothérapie ont été effectués, dans le but de déterminer si ce dernier peut être utilisé pour la dosimétrie in vivo. Ce détecteur a démontré une stabilité à court et à long terme comparable à d'autres détecteurs tels que les EPIDs également utilisés pour l'imagerie et la dosimétrie in vivo. Pour finir, une adaptation de la méthode de reconstruction de la dose a été proposée afin de pouvoir l'implémenter sur une installation de tomothérapie. Ce manuscrit est composé de deux articles et d'un script contenant des informations supplémentaires sur ce travail. Dans ce dernier, le premier chapitre introduit l'état de l'art de la dosimétrie in vivo et de la radiothérapie adaptative, et explique pourquoi nous nous intéressons à la reconstruction 3D de la dose délivrée. Dans le chapitre 2, l'algorithme 3D de calcul de dose implémenté pour ce travail est décrit, ainsi que les paramètres physiques principaux nécessaires pour le calcul de dose. Les caractéristiques du détecteur MVCT de la tomothérapie utilisé pour les mesures de transit sont décrites dans le chapitre 3. Le chapitre 4 contient un premier article intitulé '3D dose reconstruction for narrow beams using ion chamber array measurements', qui décrit la méthode de reconstruction et présente des tests de la méthodologie sur des fantômes irradiés avec des faisceaux étroits. Le chapitre 5 contient un second article intitulé 'Stability of the Helical TomoTherapy HiArt II detector for treatment beam irradiations'. Un procédé de reconstruction de la dose spécifique pour l'utilisation du détecteur MVCT de la tomothérapie est présenté au chapitre 6. Une discussion et les perspectives de la thèse de doctorat sont présentées au chapitre 7, suivies par une conclusion au chapitre 8. Le concept de la tomothérapie est exposé dans l'annexe 1. Pour finir, la radiothérapie «informationnelle 3D et la radiothérapie par modulation d'intensité sont présentées dans l'annexe 2.

Right coronary artery flow velocity and volume assessment with spiral K-space sampled breathhold velocity-encoded MRI at 3 tesla: accuracy and reproducibility.

PURPOSE: To evaluate accuracy and reproducibility of flow velocity and volume measurements in a phantom and in human coronary arteries using breathhold velocity-encoded (VE) MRI with spiral k-space sampling at 3 Tesla. MATERIALS AND METHODS: Flow velocity assessment was performed using VE MRI with spiral k-space sampling. Accuracy of VE MRI was tested in vitro at five constant flow rates. Reproducibility was investigated in 19 healthy subjects (mean age 25.4 +/- 1.2 years, 11 men) by repeated acquisition in the right coronary artery (RCA). RESULTS: MRI-measured flow rates correlated strongly with volumetric collection (Pearson correlation r = 0.99; P < 0.01). Due to limited sample resolution, VE MRI overestimated the flow rate by 47% on average when nonconstricted region-of-interest segmentation was used. Using constricted region-of-interest segmentation with lumen size equal to ground-truth luminal size, less than 13% error in flow rate was found. In vivo RCA flow velocity assessment was successful in 82% of the applied studies. High interscan, intra- and inter-observer agreement was found for almost all indices describing coronary flow velocity. Reproducibility for repeated acquisitions varied by less than 16% for peak velocity values and by less than 24% for flow volumes. CONCLUSION: 3T breathhold VE MRI with spiral k-space sampling enables accurate and reproducible assessment of RCA flow velocity.

A systolic array for linear MIMO detection based on an all-swap lattice reduction algorithm

A systolic array to implement lattice-reduction-aided lineardetection is proposed for a MIMO receiver. The lattice reductionalgorithm and the ensuing linear detections are operated in the same array, which can be hardware-efficient. All-swap lattice reduction algorithm (ASLR) is considered for the systolic design.ASLR is a variant of the LLL algorithm, which processes all lattice basis vectors within one iteration. Lattice-reduction-aided linear detection based on ASLR and LLL algorithms have very similarbit-error-rate performance, while ASLR is more time efficient inthe systolic array, especially for systems with a large number ofantennas.

Strong minimax lower bounds for learning

Minimax lower bounds for concept learning state, for example, thatfor each sample size $n$ and learning rule $g_n$, there exists a distributionof the observation $X$ and a concept $C$ to be learnt such that the expectederror of $g_n$ is at least a constant times $V/n$, where $V$ is the VC dimensionof the concept class. However, these bounds do not tell anything about therate of decrease of the error for a {\sl fixed} distribution--concept pair.\\In this paper we investigate minimax lower bounds in such a--stronger--sense.We show that for several natural $k$--parameter concept classes, includingthe class of linear halfspaces, the class of balls, the class of polyhedrawith a certain number of faces, and a class of neural networks, for any{\sl sequence} of learning rules $\{g_n\}$, there exists a fixed distributionof $X$ and a fixed concept $C$ such that the expected error is larger thana constant times $k/n$ for {\sl infinitely many n}. We also obtain suchstrong minimax lower bounds for the tail distribution of the probabilityof error, which extend the corresponding minimax lower bounds.

What kind of relationships between the evolution of pain, beliefs and bio-psycho-social complexity after a locomotor traumatism?

Introduction: Pain and beliefs have an influence on the patient's course in rehabilitation, pain causes fears and fears influence pain perception. The aim of this study is to understand pain and beliefs evolutions during rehabilitation taking into account of bio-psycho-social complexity.Patients and methods: 631 consecutive patients admitted in rehabilitation after a musculoskeletal traumatism were included and assessed at admission and at discharge. Pain was measured by VAS (Visual Analogical Scale), bio-psycho-social complexity by Intermed scale, and beliefs by judgement on Lickert scales. Four kinds of beliefs were evaluated: fear of a severe origin of pain, fear of movement, fear of pain and feeling of distress (loss of control). The association between the changes in pain and beliefs during the hospitalization was assessed by linear regressions.Results: After adjustment for gender, age, education and native language, patients with a decrease in pain during rehabilitation have higher probability of decreasing their fears. For the distress feeling, this relationship is weaker among bio-psycho-socially complex patients (odds-ratio 1.22 for each decreasing of 10mm/100 VAS) than among non-complex patients (OR 1.47). Patients with a pain decrease of 30% or more during hospitalization have higher probability of seeing their fears decrease, this relationship being stronger in complex patient for fear of a severe origin of pain.Discussion: The relationships between evolution of pain and beliefs move in the same direction. The higher a patient feels pain, the less they could be able to modify their dysfunctional beliefs. When the pain diminishes of 30% or more, the probability to challenge the beliefs is increased. The prognostic with regard to feeling of distress and fear of a severe origin of pain, is worse among bio-psycho-socially complex patients.

Model selection and error estimation

We study model selection strategies based on penalized empirical loss minimization. We point out a tight relationship between error estimation and data-based complexity penalization: any good error estimate may be converted into a data-based penalty function and the performance of the estimate is governed by the quality of the error estimate. We consider several penalty functions, involving error estimates on independent test data, empirical {\sc vc} dimension, empirical {\sc vc} entropy, andmargin-based quantities. We also consider the maximal difference between the error on the first half of the training data and the second half, and the expected maximal discrepancy, a closely related capacity estimate that can be calculated by Monte Carlo integration. Maximal discrepancy penalty functions are appealing for pattern classification problems, since their computation is equivalent to empirical risk minimization over the training data with some labels flipped.

Respostas do cacaueiro à aplicação de N, P e K em dois solos da Amazônia Brasileira

A cacauicultura da Amazônia está implantada em solos eutróficos, com predominância da Terra Roxa Estruturada, e em Latossolos ou Podzólicos distróficos, sendo desconhecidas as limitações nutricionais desses solos na fase produtiva do cacaueiro. As respostas do cacaueiro à aplicação de N, P e K foram determinadas em dois experimentos instalados nos municípios de Medicilândia, ao longo da Rodovia Transamazônica, e Benevides, Pará, em solos das unidades Terra Roxa Estrutura eutrófica (TR) e Latossolo Amarelo (LA), respectivamente. As lavouras de cacau do híbrido Sca 6 x Be 10 foram implantadas após o corte e queima da mata primária, utilizando-se, como esquema experimental, um fatorial NPK 2³ com tratamentos adicionais de P. Os resultados obtidos demonstraram que o P foi o principal nutriente que limitou a produção, provocando incrementos de produtividade (P < 0,01) da ordem de 13,7% (110 kg ha-1) e 44,3% (214 kg ha-1) nos solos TR e LA, respectivamente, na média do período 1987 a 1993. O K também aumentou (P < 0,01) o rendimento de amêndoas secas de cacau no solo LA, verificando-se, ainda, interações significativas (P < 0,05) entre N x K e P x K neste solo. A resposta linear do cacaueiro ao P e o aumento ou diminuição da taxa de infecção de vassoura-de-bruxa ao N, P e K evidenciam a necessidade de novas pesquisas para definir a dosagem econômica de P, de K e o efeito da interação dos nutrientes com a enfermidade.