Développement et validation d’un index de prédiction des risques d’institutionnalisation et de décès pour le contrôle des variables confondantes non-mesurées dans la population âgée

Le biais de confusion est un défi majeur des études observationnelles, surtout s'ils sont induits par des caractéristiques difficiles, voire impossibles, à mesurer dans les banques de données administratives de soins de santé. Un des biais de confusion souvent présents dans les études pharmacoépidémiologiques est la prescription sélective (en anglais « prescription channeling »), qui se manifeste lorsque le choix du traitement dépend de l'état de santé du patient et/ou de son expérience antérieure avec diverses options thérapeutiques. Parmi les méthodes de contrôle de ce biais, on retrouve le score de comorbidité, qui caractérise l'état de santé d'un patient à partir de médicaments délivrés ou de diagnostics médicaux rapportés dans les données de facturations des médecins. La performance des scores de comorbidité fait cependant l'objet de controverses car elle semble varier de façon importante selon la population d'intérêt. Les objectifs de cette thèse étaient de développer, valider, et comparer les performances de deux scores de comorbidité (un qui prédit le décès et l’autre qui prédit l’institutionnalisation), développés à partir des banques de services pharmaceutiques de la Régie de l'assurance-maladie du Québec (RAMQ) pour leur utilisation dans la population âgée. Cette thèse vise également à déterminer si l'inclusion de caractéristiques non rapportées ou peu valides dans les banques de données administratives (caractéristiques socio-démographiques, troubles mentaux ou du sommeil), améliore la performance des scores de comorbidité dans la population âgée. Une étude cas-témoins intra-cohorte fut réalisée. La cohorte source consistait en un échantillon aléatoire de 87 389 personnes âgées vivant à domicile, répartie en une cohorte de développement (n=61 172; 70%) et une cohorte de validation (n=26 217; 30%). Les données ont été obtenues à partir des banques de données de la RAMQ. Pour être inclus dans l’étude, les sujets devaient être âgés de 66 ans et plus, et être membres du régime public d'assurance-médicaments du Québec entre le 1er janvier 2000 et le 31 décembre 2009. Les scores ont été développés à partir de la méthode du Framingham Heart Study, et leur performance évaluée par la c-statistique et l’aire sous les courbes « Receiver Operating Curves ». Pour le dernier objectif qui est de documenter l’impact de l’ajout de variables non-mesurées ou peu valides dans les banques de données au score de comorbidité développé, une étude de cohorte prospective (2005-2008) a été réalisée. La population à l'étude, de même que les données, sont issues de l'Étude sur la Santé des Aînés (n=1 494). Les variables d'intérêt incluaient statut marital, soutien social, présence de troubles de santé mentale ainsi que troubles du sommeil. Tel que décrit dans l'article 1, le Geriatric Comorbidity Score (GCS) basé sur le décès, a été développé et a présenté une bonne performance (c-statistique=0.75; IC95% 0.73-0.78). Cette performance s'est avérée supérieure à celle du Chronic Disease Score (CDS) lorsqu'appliqué dans la population à l'étude (c-statistique du CDS : 0.47; IC 95%: 0.45-0.49). Une revue de littérature exhaustive a montré que les facteurs associés au décès étaient très différents de ceux associés à l’institutionnalisation, justifiant ainsi le développement d'un score spécifique pour prédire le risque d'institutionnalisation. La performance de ce dernier s'est avérée non statistiquement différente de celle du score de décès (c-statistique institutionnalisation : 0.79 IC95% 0.77-0.81). L'inclusion de variables non rapportées dans les banques de données administratives n'a amélioré que de 11% la performance du score de décès; le statut marital et le soutien social ayant le plus contribué à l'amélioration observée. En conclusion, de cette thèse, sont issues trois contributions majeures. D'une part, il a été démontré que la performance des scores de comorbidité basés sur le décès dépend de la population cible, d'où l'intérêt du Geriatric Comorbidity Score, qui fut développé pour la population âgée vivant à domicile. D'autre part, les médicaments associés au risque d'institutionnalisation diffèrent de ceux associés au risque de décès dans la population âgé, justifiant ainsi le développement de deux scores distincts. Cependant, les performances des deux scores sont semblables. Enfin, les résultats indiquent que, dans la population âgée, l'absence de certaines caractéristiques ne compromet pas de façon importante la performance des scores de comorbidité déterminés à partir de banques de données d'ordonnances. Par conséquent, les scores de comorbidité demeurent un outil de recherche important pour les études observationnelles.

Distributions d'auto-amorçage exactes ponctuelles des courbes ROC et des courbes de coûts

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal

How well does the ECMWF Ensemble Prediction System predict blocking?

Forecasting atmospheric blocking is one of the main problems facing medium-range weather forecasters in the extratropics. The European Centre for Medium-Range Weather Forecasts (ECMWF) Ensemble Prediction System (EPS) provides an excellent basis for medium-range forecasting as it provides a number of different possible realizations of the meteorological future. This ensemble of forecasts attempts to account for uncertainties in both the initial conditions and the model formulation. Since 18 July 2000, routine output from the EPS has included the field of potential temperature on the potential vorticity (PV) D 2 PV units (PVU) surface, the dynamical tropopause. This has enabled the objective identification of blocking using an index based on the reversal of the meridional potential-temperature gradient. A year of EPS probability forecasts of Euro-Atlantic and Pacific blocking have been produced and are assessed in this paper, concentrating on the Euro-Atlantic sector. Standard verification techniques such as Brier scores, Relative Operating Characteristic (ROC) curves and reliability diagrams are used. It is shown that Euro-Atlantic sector-blocking forecasts are skilful relative to climatology out to 10 days, and are more skilful than the deterministic control forecast at all lead times. The EPS is also more skilful than a probabilistic version of this deterministic forecast, though the difference is smaller. In addition, it is shown that the onset of a sector-blocking episode is less well predicted than its decay. As the lead time increases, the probability forecasts tend towards a model climatology with slightly less blocking than is seen in the real atmosphere. This small under-forecasting bias in the blocking forecasts is possibly related to a westerly bias in the ECMWF model. Copyright © 2003 Royal Meteorological Society

A kernel-based two-class classifier for imbalanced data sets

Many kernel classifier construction algorithms adopt classification accuracy as performance metrics in model evaluation. Moreover, equal weighting is often applied to each data sample in parameter estimation. These modeling practices often become problematic if the data sets are imbalanced. We present a kernel classifier construction algorithm using orthogonal forward selection (OFS) in order to optimize the model generalization for imbalanced two-class data sets. This kernel classifier identification algorithm is based on a new regularized orthogonal weighted least squares (ROWLS) estimator and the model selection criterion of maximal leave-one-out area under curve (LOO-AUC) of the receiver operating characteristics (ROCs). It is shown that, owing to the orthogonalization procedure, the LOO-AUC can be calculated via an analytic formula based on the new regularized orthogonal weighted least squares parameter estimator, without actually splitting the estimation data set. The proposed algorithm can achieve minimal computational expense via a set of forward recursive updating formula in searching model terms with maximal incremental LOO-AUC value. Numerical examples are used to demonstrate the efficacy of the algorithm.

FunFOLDQA: a quality assessment tool for protein-ligand binding site residue predictions

The estimation of prediction quality is important because without quality measures, it is difficult to determine the usefulness of a prediction. Currently, methods for ligand binding site residue predictions are assessed in the function prediction category of the biennial Critical Assessment of Techniques for Protein Structure Prediction (CASP) experiment, utilizing the Matthews Correlation Coefficient (MCC) and Binding-site Distance Test (BDT) metrics. However, the assessment of ligand binding site predictions using such metrics requires the availability of solved structures with bound ligands. Thus, we have developed a ligand binding site quality assessment tool, FunFOLDQA, which utilizes protein feature analysis to predict ligand binding site quality prior to the experimental solution of the protein structures and their ligand interactions. The FunFOLDQA feature scores were combined using: simple linear combinations, multiple linear regression and a neural network. The neural network produced significantly better results for correlations to both the MCC and BDT scores, according to Kendall’s τ, Spearman’s ρ and Pearson’s r correlation coefficients, when tested on both the CASP8 and CASP9 datasets. The neural network also produced the largest Area Under the Curve score (AUC) when Receiver Operator Characteristic (ROC) analysis was undertaken for the CASP8 dataset. Furthermore, the FunFOLDQA algorithm incorporating the neural network, is shown to add value to FunFOLD, when both methods are employed in combination. This results in a statistically significant improvement over all of the best server methods, the FunFOLD method (6.43%), and one of the top manual groups (FN293) tested on the CASP8 dataset. The FunFOLDQA method was also found to be competitive with the top server methods when tested on the CASP9 dataset. To the best of our knowledge, FunFOLDQA is the first attempt to develop a method that can be used to assess ligand binding site prediction quality, in the absence of experimental data.

Construction of neurofuzzy models for imbalanced data classification

We propose a new class of neurofuzzy construction algorithms with the aim of maximizing generalization capability specifically for imbalanced data classification problems based on leave-one-out (LOO) cross validation. The algorithms are in two stages, first an initial rule base is constructed based on estimating the Gaussian mixture model with analysis of variance decomposition from input data; the second stage carries out the joint weighted least squares parameter estimation and rule selection using orthogonal forward subspace selection (OFSS)procedure. We show how different LOO based rule selection criteria can be incorporated with OFSS, and advocate either maximizing the leave-one-out area under curve of the receiver operating characteristics, or maximizing the leave-one-out Fmeasure if the data sets exhibit imbalanced class distribution. Extensive comparative simulations illustrate the effectiveness of the proposed algorithms.

Comparison of signal-to-cutoff values in first, second, and third generation enzyme immunoassays for the diagnosis of HTLV-1/2 infection in ""at-risk"" individuals from Sao Paulo, Brazil

Data obtained during routine diagnosis of human T-cell lymphotropic virus type 1 (HTLV-1) and 2 (HTLV-2) in ""at-risk"" individuals from Sao Paulo, Brazil using signal-to-cutoff (S/C) values obtained by first, second, and third generation enzyme immunoassay (EIA) kits, were compared. The highest S/C values were obtained with third generation EIA kits, but no correlation was detected between these values and specific antibody reactivity to HTLV-1, HTLV-2, or untyped HTLV (p = 0.302). In addition, use of these third generation kits resulted in HTLV-1/2 false-positive samples. In contrast, first and second generation EIA kits showed high specificity, and the second generation EIA kits showed the highest efficiency, despite lower S/C values. Using first and second generation EIA kits, significant differences in specific antibody detection of HTLV-1, relative to HTLV-2 (p = 0.019 for first generation and p < 0.001 for second generation EIA kits) and relative to untyped HTLV (p = 0.025 for first generation EIA kits), were observed. These results were explained by the composition and format of the assays. In addition, using receiver operating characteristics (ROC) analysis, a slight adjustment in cutoff values for third generation EIA kits improved their specificities and should be used when HTLV ""at-risk"" populations from this geographic area are to be evaluated. (C) 2009 Elsevier B.V. All rights reserved.

A hybrid approach to learn with imbalanced classes using evolutionary algorithms

There is an increasing interest in the application of Evolutionary Algorithms (EAs) to induce classification rules. This hybrid approach can benefit areas where classical methods for rule induction have not been very successful. One example is the induction of classification rules in imbalanced domains. Imbalanced data occur when one or more classes heavily outnumber other classes. Frequently, classical machine learning (ML) classifiers are not able to learn in the presence of imbalanced data sets, inducing classification models that always predict the most numerous classes. In this work, we propose a novel hybrid approach to deal with this problem. We create several balanced data sets with all minority class cases and a random sample of majority class cases. These balanced data sets are fed to classical ML systems that produce rule sets. The rule sets are combined creating a pool of rules and an EA is used to build a classifier from this pool of rules. This hybrid approach has some advantages over undersampling, since it reduces the amount of discarded information, and some advantages over oversampling, since it avoids overfitting. The proposed approach was experimentally analysed and the experimental results show an improvement in the classification performance measured as the area under the receiver operating characteristics (ROC) curve.

Integration of Ligand- and Target-Based Virtual Screening for the Discovery of Cruzain Inhibitors

A myriad of methods are available for virtual screening of small organic compound databases. In this study we have successfully applied a quantitative model of consensus measurements, using a combination of 3D similarity searches (ROCS and EON), Hologram Quantitative Structure Activity Relationships (HQSAR) and docking (FRED, FlexX, Glide and AutoDock Vina), to retrieve cruzain inhibitors from collected databases. All methods were assessed individually and then combined in a Ligand-Based Virtual Screening (LBVS) and Target-Based Virtual Screening (TBVS) consensus scoring, using Receiving Operating Characteristic (ROC) curves to evaluate their performance. Three consensus strategies were used: scaled-rank-by-number, rank-by-rank and rank-by-vote, with the most thriving the scaled-rank-by-number strategy, considering that the stiff ROC curve appeared to be satisfactory in every way to indicate a higher enrichment power at early retrieval of active compounds from the database. The ligand-based method provided access to a robust and predictive HQSAR model that was developed to show superior discrimination between active and inactive compounds, which was also better than ROCS and EON procedures. Overall, the integration of fast computational techniques based on ligand and target structures resulted in a more efficient retrieval of cruzain inhibitors with desired pharmacological profiles that may be useful to advance the discovery of new trypanocidal agents.

Automatic spiral analysis for objective assessment of motor symptoms in Parkinson's disease

A challenge for the clinical management of advanced Parkinson’s disease (PD) patients is the emergence of fluctuations in motor performance, which represents a significant source of disability during activities of daily living of the patients. There is a lack of objective measurement of treatment effects for in-clinic and at-home use that can provide an overview of the treatment response. The objective of this paper was to develop a method for objective quantification of advanced PD motor symptoms related to off episodes and peak dose dyskinesia, using spiral data gathered by a touch screen telemetry device. More specifically, the aim was to objectively characterize motor symptoms (bradykinesia and dyskinesia), to help in automating the process of visual interpretation of movement anomalies in spirals as rated by movement disorder specialists. Digitized upper limb movement data of 65 advanced PD patients and 10 healthy (HE) subjects were recorded as they performed spiral drawing tasks on a touch screen device in their home environment settings. Several spatiotemporal features were extracted from the time series and used as inputs to machine learning methods. The methods were validated against ratings on animated spirals scored by four movement disorder specialists who visually assessed a set of kinematic features and the motor symptom. The ability of the method to discriminate between PD patients and HE subjects and the test-retest reliability of the computed scores were also evaluated. Computed scores correlated well with mean visual ratings of individual kinematic features. The best performing classifier (Multilayer Perceptron) classified the motor symptom (bradykinesia or dyskinesia) with an accuracy of 84% and area under the receiver operating characteristics curve of 0.86 in relation to visual classifications of the raters. In addition, the method provided high discriminating power when distinguishing between PD patients and HE subjects as well as had good test-retest reliability. This study demonstrated the potential of using digital spiral analysis for objective quantification of PD-specific and/or treatment-induced motor symptoms.

A Short Physical Performance Battery (SPPB) como preditora da fragilidade em idosos residentes na comunidade

Introduction: The SPPB provides information about physical function and is a predictor of adverse events in the elderly. Frailty is a multidimensional syndrome that increases susceptibility to diseases and disability. However it may be possible to prevent or postpone frailty if is identified early. Our objective is to analyze SPPB s ability in screening for frailty a community-dwelling young elderly from cities with distinct socioeconomic conditions. Methods: Data were originated from community dwelling adults (65-74 years old) in Canada (Saint Bruno; n = 60) and Brazil (Santa Cruz; n = 64). SPPB was used to assess physical performance. Frailty was defined as the presence of ≥ 3 of these criteria: weight loss, exhaustion, weakness, mobility limitation and low physical activity. One point was given for each criterion met, totalizing a frailty score ranged from 0 to 5. The Linear Regression and Receiver Operating Characteristics analyses were performed to evaluate the SPPB s screening ability. Results: Mean age was 69.48, 10.0% of the Saint Bruno s sample and 28.1% of Santa Cruz s were frail (p = 0.001), the SPPB score means were 9.6 and 8.5 respectively (p = 0.01). SPPB correlated with the frailty score (R2 = 0.33), with better results for Saint Bruno. A cutoff of 9 in SPPB had good sensitivity and specificity in discriminating frail from non frail in Saint Bruno (AUC = 0.81) but showed fair results in Santa Cruz (AUC = 0.61). Conclusion: The SPPB has moderate ability in predicting frailty among older adult s population, and is an useful test to identify people with good functionality and low frailty when SPPB scores are ≥9

Lymphocyte transformation assay for C neoformans antigen is not reliable for detecting cellular impairment in patients with Neurocryptococcosis

Background: Cryptococcus neoformans causes meningitis and disseminated infection in healthy individuals, but more commonly in hosts with defective immune responses. Cell-mediated immunity is an important component of the immune response to a great variety of infections, including yeast infections. We aimed to evaluate a specific lymphocyte transformation assay to Cryptococcus neoformans in order to identify immunodeficiency associated to neurocryptococcosis (NCC) as primary cause of the mycosis.Methods: Healthy volunteers, poultry growers, and HIV-seronegative patients with neurocryptococcosis were tested for cellular immune response. Cryptococcal meningitis was diagnosed by India ink staining of cerebrospinal fluid and cryptococcal antigen test (Immunomycol-Inc, SP, Brazil). Isolated peripheral blood mononuclear cells were stimulated with C. neoformans antigen, C. albicans antigen, and pokeweed mitogen. The amount of H-3-thymidine incorporated was assessed, and the results were expressed as stimulation index (SI) and log SI, sensitivity, specificity, and cut-off value (receiver operating characteristics curve). We applied unpaired Student t tests to compare data and considered significant differences for p<0.05.Results: The lymphotoxin alpha showed a low capacity with all the stimuli for classifying patients as responders and non-responders. Lymphotoxin alpha stimulated by heated-killed antigen from patients with neurocryptococcosis was not affected by TCD4+ cell count, and the intensity of response did not correlate with the clinical evolution of neurocryptococcosis.Conclusion: Response to lymphocyte transformation assay should be analyzed based on a normal range and using more than one stimulator. The use of a cut-off value to classify patients with neurocryptococcosis is inadequate. Statistical analysis should be based on the log transformation of SI. A more purified antigen for evaluating specific response to C. neoformans is needed.

Avaliação do desempenho diagnóstico e do valor de corte para o índice de respiração rápida e superficial na predição do insucesso da extubação

OBJETIVO: Avaliar o desempenho diagnóstico do índice de respiração rápida e superficial (IRRS) na predição do insucesso da extubação de pacientes adultos em terapia intensiva e verificar a adequação do valor de corte clássico para esse índice. MÉTODOS: Estudo prospectivo realizado na unidade de terapia intensiva de adultos do Hospital das Clínicas da Faculdade de Medicina de Botucatu, através da avaliação do IRRS em 73 pacientes consecutivos considerados clinicamente prontos para extubação. RESULTADOS: O IRRS com valor de corte clássico (105 ciclos/min/L) apresentou sensibilidade de 20% e especificidade de 95% (soma = 115%). A análise da curva receiver operator characteristic (ROC) demonstrou melhor valor de corte (76,5 ciclos/min/L), o qual forneceu sensibilidade de 66% e especificidade de 74% (soma = 140%), e a área sob a curva ROC para o IRRS foi de 0,78. CONCLUSÕES: O valor de corte clássico do IRRS se mostrou inadequado nesta casuística, prevendo apenas 20% dos pacientes com falha na extubação. A obtenção do novo valor de corte permitiu um acréscimo substancial de sensibilidade, com aceitável redução da especificidade. O valor da área sob a curva ROC indicou satisfatório poder discriminativo do índice, justificando a validação de sua aplicação.

Is urine interleukin-8 level a reliable laboratory test for diagnosing late onset sepsis in premature infants?

The present study is aimed to determine serum and urine interleukin-8 (IL-8) levels in premature infants with late onset sepsis (LOS) and to evaluate if urine IL-8 is a useful test for LOS diagnosis. Fifty-six premature infants admitted to the NICU over 1 year had serum and urine IL-8 determined by ELISA. They were divided into three groups: I definite sepsis, II probable sepsis and III non-infected. Results were expressed as mean or median. Differences between groups were assessed by ANOVA, Kruskal-Wallis ANOVA and Dunns Method. Sensitivity, specificity and positive and negative predictive values were calculated and a receiver operator characteristic curve was constructed to determine serum and urine IL-8 accuracy. There were no differences between groups for birth weight, and gestational and post-natal age. Median serum and urine IL-8 levels were significantly higher in GI and GII: 929 x 906 x 625pg/ml; P=0.024, and 249 x 189 x 42pg/mgCr; P< 0.001. Optimal cut-off point was 625pg/ml for serum IL-8 with 69 sensitivity and 75pg/mgCr for urine IL-8 with 92 sensitivity. IL-8 can be determined in urine from premature infants with LOS and is an accurate and feasible diagnosis method.

Accuracy of sagittal abdominal diameter as predictor of abdominal fat among Brazilian adults: a comparation with waist circumference

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)