Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Thematic program of Patient-related and Public Health Research Doctoral seminar Modeling agreement on a bounded scale Date: 31 May 2016 at 3 pm Location and room: live in meeting room A (02.53), L-BioStat, KU Leuven and via video in room E139, UHasselt Speaker: Sophie Vanbelle, Maastricht University, the Netherlands Abstract Agreement is an important concept in the development process of new measurement instruments. For example, when a true gold standard is not available, the agreement between the new instrument and an established one is used to determine criterion validity. Agreement is also important in clinical decision making where disagreements possibly imply a different management of the patient. For quantitative scales, the concordance correlation coefficient (CCC) is the appropriate measure to quantify agreement between two scorers. This measure implicitly assumes that the joint distribution of the scores is bivariate normal, which is often not the case for bounded outcome scores. Bounded outcome scores are common in medical and behavioral sciences. Typical examples are visual analog scales (VAS) and scores computed as the number of positive items on a questionnaire. These kinds of scores often show a non-standard distribution, like a J- or U-shape. The Logit-Normal (LN) distribution has shown to be successful in modeling bounded outcome scores with non-standard distributions. In particular, Lesaffre et al. (2007) developed a methodology to handle bounded scores of two types: (1) when the bounded score is a coarsened version of a latent score with a LN distribution on the [0,1] interval, like VAS scales and (2) when the bounded score is a proportion with the true probability having a LN distribution. In the present work, a model-based approach, based on a bivariate generalization of the LN distribution, is developed in a Bayesian framework. This method permits to directly study the impact of predictors on the CCC and can be simply implemented in standard softwares, like JAGS and WINBUGS. The performances of new method will be compared to the classical approach for various non-standard distributions of bounded outcomes scores using simulations. Finally, three studies will illustrate the methodology: (1) a study on hypertension, where the effect of the complexity of the drug regimen on the agreement between the patient and the physician assessment of compliance (100mm VAS scale) is investigated, (2) a study on rheumatoid arthritis, where the agreement between the percentage of joints with synovitis detected using a PET/CT scan and ultrasonography is determined and (3) the Tandmobiel study, where the agreement between the child and the mother assessment of the importance of having a healthy mouth (10-points scale) is of interest. Lesaffre E., Rizopoulos D. and Tsonaka, R. (2007). The logistic transform for bounded outcome scores. Biostatistics, 8: 72-85 SEMINAR Date: 3 June at 3 pm Location and room: live in meeting room A (02.53), L-BioStat, KU Leuven and via video in room E139, UHasselt Speaker: Kazem Nasserinejad, Erasmus MC Rotterdam, the Netherlands Title: Prediction of hemoglobin in blood donors using a latent class mixed-effects transition model and proposing a criteria for choosing the number of latent classes in Bayesian setting Abstract Blood donors experience a temporary reduction in their hemoglobin (Hb) value after donation. Due to recovery process after each donation as well as state dependence and unobserved heterogeneity, longitudinal data of Hb values of blood donors provide unique statistical challenges. To model recovery process after donation a latent-class mixed-effects transition model with a flexible function was proposed. The latent classes identify groups of donors with fast or slow recovery times and donors whose recovery time increases with the number of donations. The transition effect accounts for possible state dependence in the observed data. This model was estimated in a Bayesian setting with some informative priors for parameters of the recovery process that were not identified using the observed data. Identifying the number of classes in Bayesian finite mixture models is a challenging problem. Several criteria have been proposed. It was recently shown that in overfitted mixture models, the overfitted latent classes will asymptotically become empty under specific conditions for the prior of the class proportions. This result may be used to construct a criterion for finding the true number of latent classes, based on the removal of latent classes that have negligible proportions. This criterion can easily be implemented in complex statistical models using standard Bayesian software. We performed an extensive simulation study to develop practical guidelines to determine the appropriate number of latent classes based on the posterior distribution of the class proportions, and to compare this criterion with alternative criteria. The performance of the proposed criterion is illustrated using hemoglobin values of blood donors. Prediction of hemoglobin in blood donors using a latent class mixed-effects transition model and proposing a criteria for choosing the number of latent classes in Bayesian setting