Nutrimetabolomics fingerprinting to identify biomarkers of bread exposure in a free-living population from the PREDIMED study cohort

Bread is one of the most widely consumed foods. Its impact on human health is currently of special interest for researchers. We aimed to identify biomarkers of bread consumption by applying a nutrimetabolomic approach to a free-living population. An untargeted HPLC q-TOF-MS and multivariate analysis was applied to human urine from 155 subjects stratified by habitual bread consumption in three groups: non-consumers of bread (n = 56), white-bread consumers (n = 48) and whole-grain bread consumers (n = 51). The most differential metabolites (variable importance for projection ≥1.5) included compounds originating from cereal plant phytochemicals such as benzoxazinoids and alkylresorcinol metabolites, and compounds produced by gut microbiota (such as enterolactones, hydroxybenzoic and dihydroferulic acid metabolites). Pyrraline, riboflavin, 3-indolecarboxylic acid glucuronide, 2,8-dihydroxyquinoline glucuronide and N-α-acetylcitrulline were also tentatively identified. In order to combine multiple metabolites in a model to predict bread consumption, a stepwise logistic regression analysis was used. Receiver operating curves were constructed to evaluate the global performance of individual metabolites and their combination. The area under the curve values [AUC (95 % CI)] of combined models ranged from 77.8 % (69.1 86.4 %) to 93.7 % (89.4 98.1 %), whereas the AUC for the metabolites included in the models had weak values when they were evaluated individually: from 58.1 % (46.6 69.7 %) to 78.4 % (69.8 87.1 %). Our study showed that a daily bread intake significantly impacted on the urinary metabolome, despite being examined under uncontrolled free-living conditions. We further concluded that a combination of several biomarkers of exposure is better than a single biomarker for the predictive ability of discriminative analysis.

Eye Fundus Image Analysis for Automatic Detection of Diabetic Retinopathy

Diabetes is a rapidly increasing worldwide problem which is characterised by defective metabolism of glucose that causes long-term dysfunction and failure of various organs. The most common complication of diabetes is diabetic retinopathy (DR), which is one of the primary causes of blindness and visual impairment in adults. The rapid increase of diabetes pushes the limits of the current DR screening capabilities for which the digital imaging of the eye fundus (retinal imaging), and automatic or semi-automatic image analysis algorithms provide a potential solution. In this work, the use of colour in the detection of diabetic retinopathy is statistically studied using a supervised algorithm based on one-class classification and Gaussian mixture model estimation. The presented algorithm distinguishes a certain diabetic lesion type from all other possible objects in eye fundus images by only estimating the probability density function of that certain lesion type. For the training and ground truth estimation, the algorithm combines manual annotations of several experts for which the best practices were experimentally selected. By assessing the algorithm’s performance while conducting experiments with the colour space selection, both illuminance and colour correction, and background class information, the use of colour in the detection of diabetic retinopathy was quantitatively evaluated. Another contribution of this work is the benchmarking framework for eye fundus image analysis algorithms needed for the development of the automatic DR detection algorithms. The benchmarking framework provides guidelines on how to construct a benchmarking database that comprises true patient images, ground truth, and an evaluation protocol. The evaluation is based on the standard receiver operating characteristics analysis and it follows the medical practice in the decision making providing protocols for image- and pixel-based evaluations. During the work, two public medical image databases with ground truth were published: DIARETDB0 and DIARETDB1. The framework, DR databases and the final algorithm, are made public in the web to set the baseline results for automatic detection of diabetic retinopathy. Although deviating from the general context of the thesis, a simple and effective optic disc localisation method is presented. The optic disc localisation is discussed, since normal eye fundus structures are fundamental in the characterisation of DR.

Carga viral do papilomavirus humano na predição da gravidade de lesões cervicais em mulheres com atipias celulares na colpocitologia oncológica

OBJETIVO: avaliar o desempenho da carga viral do HPV por captura de híbridos II (CHII) na predição da gravidade das lesões cervicais. MÉTODOS: foram incluídas 309 mulheres admitidas por resultado anormal da colpocitologia oncológica (CO) entre agosto de 200 e novembro de 2002. Todas foram submetidas a avaliação histológica, sendo que a presença de neoplasia intra-epitelial cervical (NIC) grau 2 ou mais (NIC 3, carcinoma invasor) foi considerada doença grave. A CHII foi realizada para tipos de HPV de alto risco oncogênico e a carga viral medida em unidades relativas de luz (URL). O desempenho da CHII foi avaliado por curva receiver operating characteristics (ROC). RESULTADOS: na avaliação histológica, 140 (45,3%) mulheres apresentavam cervicite ou NIC 1 e 199 (54,7%), NIC 2/3, adenocarcinoma in situ ou câncer invasor. O melhor ponto de corte da CHII para a detecção de doença grave foi 35 URL, com sensibilidade de 69% e especificidade de 70%. O valor preditivo positivo das alterações compatíveis com lesão de alto grau na CO associado a CHII de 35 URL (unidades relativas de luz) foi de 88,2% para a detecção de NIC 2 ou mais. Já 95,7% das mulheres com lesões de baixo grau na CO e CHII menor que 1 URL não apresentaram lesões histológicas graves. CONCLUSÃO: o melhor desempenho da CHII no diagnóstico de NIC 2 ou lesão mais grave foi encontrado com 35 URL. A associação da CO com a CHII em diferentes cargas virais mostrou valores preditivos positivos e negativos muito altos.

Dopplerfluxometria de ducto venoso: identificação não invasiva da acidemia em fetos prematuros centralizados

OBJETIVO: estabelecer o ponto de corte a partir do qual seja possível identificar fetos prematuros com centralização do fluxo sangüíneo que apresentem gasometria anormal. MÉTODO: foi realizado estudo observacional transversal, cuja população consistia de 60 gestantes com fetos centralizados (relação umbílico-cerebral maior que 1), com idade gestacional entre 25 e 33 semanas. O ducto venoso foi identificado com auxílio da dopplerfluxometria colorida e obtida a relação S/A a partir do sonograma (relação entre a velocidade de pico da sístole ventricular e a velocidade de pico da sístole atrial). Imediatamente após a cesariana foi colhida amostra de sangue da veia umbilical para gasometria. Os conceptos foram classificados de acordo com a análise gasométrica e considerados anormais quando pH <7,20 e BE < -6 mmol/l. Após o cálculo da sensibilidade e especificidade para cada ponto de corte foi construída a curva ROC (receiver operator characteristic). RESULTADOS: foram selecionadas 60 gestantes no período de janeiro de 1998 a janeiro de 2003. No momento do estudo a idade gestacional variou entre 25 e 33 semanas, média de 29,7 semanas (±1,8 semanas). Todos os fetos apresentavam-se centralizados; dentre eles 14 apresentavam gasometria anormal ao nascimento e 46 apresentavam gasometria normal. A prevalência de resultados de fetos com gasometria anormal no material estudado foi de 23,33%. Observou-se associação significativa entre o resultado anormal do ducto venoso e gasometria anormal ao nascimento (chi2 = 784,44, p < 0,00001). O ponto de corte da relação S/A (onde a curva ROC "muda de tendência") foi 3,4. CONCLUSÃO: a análise da relação S/A do ducto venoso se mostra adequada para a predição não invasiva de gasometria anormal em fetos prematuros, centralizados, quando os valores são superiores a 3,4.

Diagnóstico da restrição de crescimento fetal pela relação diâmetro transverso do cerebelo e circunferência abdominal

OBJETIVO: testar a validade da relação diâmetro transverso do cerebelo (DTC)/circunferência abdominal (CA) no diagnóstico da restrição de crescimento fetal (RCF), determinando seu melhor ponto de corte e acurácia nas restrições simétrica e assimétrica. MÉTODOS: estudo prospectivo, transversal, envolvendo 250 gestantes com gravidez única, idade gestacional entre a 20ª e a 42ª semana, confirmada por ultra-sonografia. A medida do DTC foi obtida colocando-se os calipers nas margens externas do cerebelo, após sua localização na fossa posterior, com suave rotação do transdutor abaixo do plano do tálamo. A circunferência abdominal foi medida na junção das veias porta esquerda e umbilical. O melhor ponto de corte da relação DTC/CA foi obtido pela curva ROC (receiver operator characteristic). Os neonatos cujas relações DTC/CA foram maiores do que o ponto de corte selecionado foram considerados com RCF. Consideraram-se padrão-ouro para o diagnóstico de RCF recém-nascidos cujos pesos situaram-se abaixo do percentil 10. Neonatos com RCF e índice ponderal de Rohrer entre 2,2 e 3 foram considerados simétricos e abaixo de 2,2, assimétricos. RESULTADOS: o ponto de corte da relação DTC/CA, obtido pela curva ROC foi 16,1. A sensibilidade, especificidade, acurácia, valores preditivos positivo e negativo e razões de verossimilhança positiva e negativa foram de 77,4; 82,6; 38,7; 96,3; 82; 4,5 e 3,7%, respectivamente. Na RCF simétrica a sensibilidade e especificidade foram de 80,8 e 81,7%, respectivamente. Na assimétrica, a sensibilidade e especificidade foram de 60 e 75%, respectivamente. CONCLUSÃO: a relação DTC/CA mostrou-se eficaz no diagnóstico da RCF simétrica e assimétrica.

Punção aspirativa de linfonodo axilar guiada pela ultrassonografia é eficaz como método de predição de acometimento linfonodal em pacientes com câncer de mama?

OBJETIVO: Avaliar a viabilidade e acurácia diagnóstica da ultrassonografia pré-operatória combinada com biopsia por agulha fina (US-PAAF) e do exame clínico da axila em pacientes com câncer de mama. MÉTODOS: Neste estudo prospectivo 171 axilas de pacientes com câncer de mama foram avaliadas pelo exame clínico e ultrassonografia (US) com e sem biopsia por agulha fina (PAAF). Os linfonodos com espessura cortical maior que 2,3 mm na ultrassonografia foram considerados suspeitos e submetidos a US-PAAF. RESULTADOS: A análise de regressão logística não mostrou correlação estatisticamente significativa entre exame clínico e axilas positivas no exame patológico. Em relação à avaliação axilar com US, o risco de achados anatomopatológicos positivos aumentou 12,6 vezes, valor Kappa de Cohen foi de 0,12 para exame clínico, 0,48 para US e 0,80 para US-PAAF. A acurácia foi de 61,4% para o exame clínico, 73,1% para os US e 90,1% para US-PAAF. Análise Receiver Operating Chracteristics (ROC) mostrou que uma espessura de 2,75 mm cortical correspondeu à mais elevada sensibilidade e especificidade na predição metástase axilar (82,7 e 82,2%, respectivamente). CONCLUSÕES: A US combinada com aspiração por agulha fina é mais precisa que o exame clínico na avaliação do status axilar no pré-operatório em mulheres com câncer de mama. Aquelas que são US-PAAF positivo podem ser direcionadas para esvaziamento linfonodal axilar imediatamente, e somente aqueles que são US-PAAF negativos devem ser considerados para biópsia de linfonodo sentinela.

Linguistic models for decision support

Linguistic modelling is a rather new branch of mathematics that is still undergoing rapid development. It is closely related to fuzzy set theory and fuzzy logic, but knowledge and experience from other fields of mathematics, as well as other fields of science including linguistics and behavioral sciences, is also necessary to build appropriate mathematical models. This topic has received considerable attention as it provides tools for mathematical representation of the most common means of human communication - natural language. Adding a natural language level to mathematical models can provide an interface between the mathematical representation of the modelled system and the user of the model - one that is sufficiently easy to use and understand, but yet conveys all the information necessary to avoid misinterpretations. It is, however, not a trivial task and the link between the linguistic and computational level of such models has to be established and maintained properly during the whole modelling process. In this thesis, we focus on the relationship between the linguistic and the mathematical level of decision support models. We discuss several important issues concerning the mathematical representation of meaning of linguistic expressions, their transformation into the language of mathematics and the retranslation of mathematical outputs back into natural language. In the first part of the thesis, our view of the linguistic modelling for decision support is presented and the main guidelines for building linguistic models for real-life decision support that are the basis of our modeling methodology are outlined. From the theoretical point of view, the issues of representation of meaning of linguistic terms, computations with these representations and the retranslation process back into the linguistic level (linguistic approximation) are studied in this part of the thesis. We focus on the reasonability of operations with the meanings of linguistic terms, the correspondence of the linguistic and mathematical level of the models and on proper presentation of appropriate outputs. We also discuss several issues concerning the ethical aspects of decision support - particularly the loss of meaning due to the transformation of mathematical outputs into natural language and the issue or responsibility for the final decisions. In the second part several case studies of real-life problems are presented. These provide background and necessary context and motivation for the mathematical results and models presented in this part. A linguistic decision support model for disaster management is presented here – formulated as a fuzzy linear programming problem and a heuristic solution to it is proposed. Uncertainty of outputs, expert knowledge concerning disaster response practice and the necessity of obtaining outputs that are easy to interpret (and available in very short time) are reflected in the design of the model. Saaty’s analytic hierarchy process (AHP) is considered in two case studies - first in the context of the evaluation of works of art, where a weak consistency condition is introduced and an adaptation of AHP for large matrices of preference intensities is presented. The second AHP case-study deals with the fuzzified version of AHP and its use for evaluation purposes – particularly the integration of peer-review into the evaluation of R&D outputs is considered. In the context of HR management, we present a fuzzy rule based evaluation model (academic faculty evaluation is considered) constructed to provide outputs that do not require linguistic approximation and are easily transformed into graphical information. This is achieved by designing a specific form of fuzzy inference. Finally the last case study is from the area of humanities - psychological diagnostics is considered and a linguistic fuzzy model for the interpretation of outputs of multidimensional questionnaires is suggested. The issue of the quality of data in mathematical classification models is also studied here. A modification of the receiver operating characteristics (ROC) method is presented to reflect variable quality of data instances in the validation set during classifier performance assessment. Twelve publications on which the author participated are appended as a third part of this thesis. These summarize the mathematical results and provide a closer insight into the issues of the practicalapplications that are considered in the second part of the thesis.

In vitro-induced antibody production in chronic hepatitis C virus infection

The objectives of the present study were to assess the in vitro-induced anti-hepatitis C virus (HCV) antibody production (IVIAP) in relation to the clinical, biochemical, virologic and histologic variables of patients with HCV infection. The study included 57 patients (60% males) with HCV infection (anti-HCV and HCV-RNA positive). Alanine aminotransferase (ALT) was elevated in 89% of the patients. Mean viral load was 542,241 copies/ml and histology of the liver showed chronic hepatitis in 27/52 (52%) and cirrhosis in 11/52 (21%) patients. IVIAP levels were determined by immunoenzymatic assay at median absorbance of 0.781 at 450 nm. IVIAP was negative in 14% of the patients. When groups with IVIAP levels above and below the median were compared, high IVIAP levels were associated with the male sex, elevated ALT levels and more advanced disease stage. After logistic regression analysis, advanced histologic damage to the liver remained as the only independent variable associated with elevated IVIAP levels. Using a receiver operator characteristic curve, the best cut-off level for IVIAP was established (= 1.540), with 71% sensitivity and 94% specificity for the detection of more advanced disease stages (grades 3 and 4). These findings are consistent with the participation of immunological mechanisms in the genesis of the hepatic lesions induced by HCV and indicate that the IVIAP test may be useful as a noninvasive marker of liver damage either alone or in combination with other markers.

An analysis of the electrocardiogram QT interval

No reports testing the efficacy of the use of the QT/RR ratio <1/2 for detecting a normal QTc interval were found in the literature. The objective of the present study was to determine if a QT/RR ratio <=1/2 can be considered to be equal to the normal QTc and to compare the QT and QTc measured and calculated clinically and by a computerized electrocardiograph. Ratios (140 QT/RR) of 28 successive electrocardiograms obtained from 28 consecutive patients in a tertiary level teaching hospital were analyzed clinically by 5 independent observers and by a computerized electrocardiograph. The QT/RR ratio provided 56% sensitivity and 78% specificity, with an area under the receiver operator characteristic curve of 75.8% (95%CI: 0.68 to 0.84). The divergence in QT and QTc interval measurements between clinical and computerized evaluation were 0.01 ± 0.03 s (95%CI: 0.04-0.02) and 0.01 ± 0.04 s (95%CI: -0.05-0.03), respectively. The QT and QTc values measured clinically and by a computerized electrocardiograph were similar. The QT/RR ratio <=1/2 was not a satisfactory index for QTc evaluation because it could not predict a normal QTc value.

Efficacy of the FEV1/FEV6 ratio compared to the FEV1/FVC ratio for the diagnosis of airway obstruction in subjects aged 40 years or over

The present cross-sectional, population-based study was designed to evaluate the performance of the FEV1/FEV6 ratio for the detection of airway-obstructed subjects compared to the FEV1/FVC <0.70 fixed ratio test, as well as the lower limit of normality (LLN) for 1000 subjects ³40 years of age in the metropolitan area of São Paulo, SP, Brazil. After the exclusion of 37 (3.7%) spirometries, a total of 963 pre-bronchodilator (BD) and 918 post-BD curves were constructed. The majority of the post-BD curves (93.1%) were of very good quality and achieved grade A (762 curves) or B (93 curves). The FEV1/FEV6 and FEV1/FVC ratios were highly correlated (r² = 0.92, P < 0.000). Two receiver operator characteristic curves were constructed in order to express the imbalance between the sensitivity and specificity of the FEV1/FEV6 ratio compared to two FEV1/FVC cut-off points for airway obstruction: equal to 70 (area under the curve = 0.98, P < 0.0001) and the LLN (area under the curve = 0.97, P < 0.0001), in the post-BD curves. According to an FEV1/FVC <0.70, the cut-off point for the FEV1/FEV6 ratio with the highest sum for sensitivity and specificity was 0.75. The FEV1/FEV6 ratio can be considered to be a good alternative to the FEV1/FVC ratio for the diagnosis of airway obstruction, both using a fixed cut-off point or below the LLN as reference. The FEV1/FEV6 ratio has the additional advantage of being an easier maneuver for the subjects and for the lung function technicians, providing a higher reproducibility than traditional spirometry maneuvers.

Evaluation of a 1-h 75-g oral glucose tolerance test in the diagnosis of gestational diabetes

In order to evaluate the performance of a 1-h 75-g oral glucose tolerance test (OGTT) for the diagnosis of gestational diabetes mellitus (GDM), a cohort of 4998 women, 20 years or older, without previous diabetes being treated in prenatal care clinics in Brazil answered a questionnaire and performed a 75-g OGTT including fasting, 1-h and 2-h glucose measurements between their 24th and 28th gestational weeks. Pregnancy outcomes were transcribed from medical registries. GDM was defined according to WHO criteria (fasting: ≥126 mg/dL; 2-h value: ≥140 mg/dL) and macrosomia as a birth weight equal to or higher than 4000 g. Areas under the receiver operator characteristic curve (AUC) were compared and diagnostic properties of various cut-off points were evaluated. The AUCs for the prediction of macrosomia were 0.606 (0.572-0.637) for the 1-h and 0.589 (0.557-0.622) for the 2-h plasma glucose test. Similar predictability was demonstrable regarding combined adverse outcomes: 0.582 (0.559-0.604) for the 1-h test and 0.572 (0.549-0.595) for the 2-h test. When the 1-h glucose test was evaluated against a diagnosis of GDM defined by the 2-h glucose test, the AUC was 0.903 (0.886-0.919). The cut-off point that maximized sensitivity (83%) and specificity (83%) was 141 mg/dL, identifying 21% of the women as positive. A cut-off point of 160 mg/dL, with lower sensitivity (62%), had higher specificity (94%), labeling 8.6% as positive. Detection of GDM can be done with a 1-h 75-g OGTT: the value of 160 mg/dL has the same diagnostic performance as the conventional 2-h value (140 mg/dL). The simplification of the test may improve coverage and timing of the diagnosis of GDM.

Comparison of non-HDL cholesterol and waist circumference vs. triglyceride levels and waist circumference in predicting coronary heart disease risk

BACKGROUND: Dyslipidemia is recognized as a major cause of coronary heart disease (CHD). Emerged evidence suggests that the combination of triglycerides (TG) and waist circumference can be used to predict the risk of CHD. However, considering the known limitations of TG, non-high-density lipoprotein (non-HDL = Total cholesterol - HDL cholesterol) cholesterol and waist circumference model may be a better predictor of CHD. PURPOSE: The Framingham Offspring Study data were used to determine if combined non-HDL cholesterol and waist circumference is equivalent to or better than TG and waist circumference (hypertriglyceridemic waist phenotype) in predicting risk of CHD. METHODS: A total of3,196 individuals from Framingham Offspring Study, aged ~ 40 years old, who fasted overnight for ~ 9 hours, and had no missing information on nonHDL cholesterol, TG levels, and waist circumference measurements, were included in the analysis. Receiver Operator Characteristic Curve (ROC) Area Under the Curve (AUC) was used to compare the predictive ability of non-HDL cholesterol and waist circumference and TG and waist circumference. Cox proportional-hazards models were used to examine the association between the joint distributions of non-HDL cholesterol, waist circumference, and non-fatal CHD; TG, waist circumference, and non-fatal CHD; and the joint distribution of non-HDL cholesterol and TG by waist circumference strata, after adjusting for age, gender, smoking, alcohol consumption, diabetes, and hypertension status. RESULTS: The ROC AUC associated with non-HDL cholesterol and waist circumference and TG and waist circumference are 0.6428 (CI: 0.6183, 0.6673) and 0.6299 (CI: 0.6049, 0.6548) respectively. The difference in the ROC AVC is 1.29%. The p-value testing if the difference in the ROC AVCs between the two models is zero is 0.10. There was a strong positive association between non-HDL cholesterol and the risk for non-fatal CHD within each TO levels than that for TO levels within each level of nonHDL cholesterol, especially in individuals with high waist circumference status. CONCLUSION: The results suggest that the model including non-HDL cholesterol and waist circumference may be superior at predicting CHD compared to the model including TO and waist circumference.

The relationship between body mass index and breast cancer recurrence/progression and breast cancer-specific death

The aim of this study was to describe the nonlinear association between body mass index (BMI) and breast cancer outcomes and to determine whether BMI improves prediction of outcomes. A cohort of906 breast cancer patients diagnosed at Henry Ford Health System, Detroit (1985-1990) were studied. The median follow-up was 10 years. Multivariate logistic regression was used to model breast cancer recurrence/progression and breast cancer-specific death. Restricted cubic splines were used to model nonlinear effects. Receiver operator characteristic areas under the curves (ROC AUC) were used to evaluate prediction. BMI was nonlinearly associated with recurrence/progression and death (p= 0.0230 and 0.0101). Probability of outcomes increased with increase or decrease ofBMI away from 25. BMI splines were suggestive of improved prediction of death. The ROC AUCs for nested models with and without BMI were 0.8424 and 0.8331 (p= 0.08). I f causally associated, modifying patients BMI towards 25 may improve outcomes.

THE PREVALENCE AND PREDICTION OF PULMONARY FIBROSIS IN AN

This thesis describes an ancillary project to the Early Diagnosis of Mesothelioma and Lung Cancer in Prior Asbestos Workers study and was conducted to determine the effects of asbestos exposure, pulmonary function and cigarette smoking in the prediction of pulmonary fibrosis. 613 workers who were occupationally exposed to asbestos for an average of 25.9 (SD=14.69) years were sampled from Sarnia, Ontario. A structured questionnaire was administered during a face-to-face interview along with a low-dose computed tomography (LDCT) of the thorax. Of them, 65 workers (10.7%, 95%CI 8.12—12.24) had LDCT-detected pulmonary fibrosis. The model predicting fibrosis included the variables age, smoking (dichotomized), post FVC % splines and post- FEV1% splines. This model had a receiver operator characteristic area under the curve of 0.738. The calibration of the model was evaluated with R statistical program and the bootstrap optimism-corrected calibration slope was 0.692. Thus, our model demonstrated moderate predictive performance.

Évaluation d’un prototype de détecteur de glucose dans le tissu interstitiel sans aiguille, le PGS (Photonic Glucose Sensor)

Objectif : Déterminer la fiabilité et la précision d’un prototype d’appareil non invasif de mesure de glucose dans le tissu interstitiel, le PGS (Photonic Glucose Sensor), en utilisant des clamps glycémiques multi-étagés. Méthodes : Le PGS a été évalué chez 13 sujets avec diabète de type 1. Deux PGS étaient testés par sujet, un sur chacun des triceps, pour évaluer la sensibilité, la spécificité, la reproductibilité et la précision comparativement à la technique de référence (le Beckman®). Chaque sujet était soumis à un clamp de glucose multi-étagé de 8 heures aux concentrations de 3, 5, 8 et 12 mmol/L, de 2 heures chacun. Résultats : La corrélation entre le PGS et le Beckman® était de 0,70. Pour la détection des hypoglycémies, la sensibilité était de 63,4%, la spécificité de 91,6%, la valeur prédictive positive (VPP) 71,8% et la valeur prédictive négative (VPN) 88,2%. Pour la détection de l’hyperglycémie, la sensibilité était de 64,7% et la spécificité de 92%, la VPP 70,8% et la VPN : 89,7%. La courbe ROC (Receiver Operating Characteristics) démontrait une précision de 0,86 pour l’hypoglycémie et de 0,87 pour l’hyperglycémie. La reproductibilité selon la « Clark Error Grid » était de 88% (A+B). Conclusion : La performance du PGS était comparable, sinon meilleure que les autres appareils sur le marché(Freestyle® Navigator, Medtronic Guardian® RT, Dexcom® STS-7) avec l’avantage qu’il n’y a pas d’aiguille. Il s’agit donc d’un appareil avec beaucoup de potentiel comme outil pour faciliter le monitoring au cours du traitement intensif du diabète. Mot clés : Diabète, diabète de type 1, PGS (Photonic Glucose Sensor), mesure continue de glucose, courbe ROC, « Clark Error Grid».