Bayesian hierarchical models and prior elicitation for fitting psychometric functions.

Our previous articles demonstrated how to analyze psychophysical data from a group of participants using generalized linear mixed models (GLMM) and two-level methods. The aim of this article is to revisit hierarchical models in a Bayesian framework. Bayesian models have been previously discussed for the analysis of psychometric functions although this approach is still seldom applied. The main advantage of using Bayesian models is that if the prior is informative, the uncertainty of the parameters is reduced through the combination of prior knowledge and the experimental data. Here, we evaluate uncertainties between and within participants through posterior distributions. To demonstrate the Bayesian approach, we re-analyzed data from two of our previous studies on the tactile discrimination of speed. We considered different methods to include a priori knowledge in the prior distribution, not only from the literature but also from previous experiments. A special type of Bayesian model, the power prior distribution, allowed us to modulate the weight of the prior, constructed from a first set of data, and use it to fit a second one. Bayesian models estimated the probability distributions of the parameters of interest that convey information about the effects of the experimental variables, their uncertainty, and the reliability of individual participants. We implemented these models using the software Just Another Gibbs Sampler (JAGS) that we interfaced with R with the package rjags. The Bayesian hierarchical model will provide a promising and powerful method for the analysis of psychometric functions in psychophysical experiments.

Combining individual and aggregated data to investigate the role of socioeconomic disparities on cancer burden in Italy.

Quantifying socioeconomic disparities and understanding the roots of inequalities are growing topics in cancer research. However, socioeconomic differences are challenging to investigate mainly due to the lack of accurate data at individual-level, while aggregate indicators are only partially informative. We implemented a multiple imputation algorithm within a statistical matching framework that combines diverse sources of data to estimate individual-level associations between income and risk of breast and lung cancer, adjusting for potential confounding factors in Italy. The framework is computationally flexible and can be adapted to similar contexts.

Development, validity and reliability of the Italian version of the Copenhagen neck functional disability scale.

BACKGROUND: Valid and reliable patient-reported outcome measures support health professionals in evaluating the results of clinical research and practice. The Copenhagen Neck Functional Disability Scale (CNFDS) has shown promising measurement properties to measure disability in patients with neck pain, but an Italian version of this questionnaire is not available. The objective of this study was to cross-culturally adapt the CNFDS into Italian (CNFDS-I), and to assess its validity and reliability in patients with neck pain. METHODS: The CNFDS-I was developed according to well-established guidelines for cross-cultural adaptation of patient-reported outcome measures. A cross-sectional clinimetric study was conducted to evaluate its validity and reliability. Patients with chronic neck pain (pain > 3 months) participated in this study. The following measurement properties (defined by the COSMIN initiative) were assessed: structural validity (exploratory factor analysis), internal consistency (Cronbach's α), construct validity [by testing hypotheses on expected correlations with the Neck Disability Index (NDI), the Neck Bournemouth Questionnaire (NBQ), and pain Visual Analogue Scale (VAS)]. Test-retest reliability [Intraclass Correlation Coefficient for agreement (ICCagreement)], and measurement error [Smallest Detectable Change (SDC)] were also assessed in 50 clinically stable patients. Floor/ceiling effects and acceptability were calculated. RESULTS: One-hundred and sixty-two patients (mean age = 47.9 ± 14.5 years, 70% female) were included. The CNFDS-I exhibited sufficient unidimensionality (one factor explained 83% of the variability) and internal consistency (α = 0.83). Construct validity was sufficient as all correlations with the other questionnaires were as expected (r = 0.846 with NDI, r = 0.708 with NBQ, r = 0.570 with VAS). Test-retest reliability was excellent (ICCagreement = 0.99, 95% CI from 0.995 to 0.999), while measurement error was equal to 8.31 scale points (27% scale range). No floor/ceiling effects were detected. The average time for filling the questionnaire was two minutes. CONCLUSIONS: The CNFDS-I proved to be a valid and reliable outcome measure to assess disability in patients with chronic neck pain. Head-to-head comparison studies on the CNFDS-I measurement properties against other disability measures for neck pain (e.g. NDI and NBQ) are required to determine the relative merits of these different measures.

Rolling Motion Along an Incline: Visual Sensitivity to the Relation Between Acceleration and Slope.

People easily intercept a ball rolling down an incline, despite its acceleration varies with the slope in a complex manner. Apparently, however, they are poor at detecting anomalies when asked to judge artificial animations of descending motion. Since the perceptual deficiencies have been reported in studies involving a limited visual context, here we tested the hypothesis that judgments of naturalness of rolling motion are consistent with physics when the visual scene incorporates sufficient cues about environmental reference and metric scale, roughly comparable to those present when intercepting a ball. Participants viewed a sphere rolling down an incline located in the median sagittal plane, presented in 3D wide-field virtual reality. In different experiments, either the slope of the plane or the sphere acceleration were changed in arbitrary combinations, resulting in a kinematics that was either consistent or inconsistent with physics. In Experiment 1 (slope adjustment), participants were asked to modify the slope angle until the resulting motion looked natural for a given ball acceleration. In Experiment 2 (acceleration adjustment), instead, they were asked to modify the acceleration until the motion on a given slope looked natural. No feedback about performance was provided. For both experiments, we found that participants were rather accurate at finding the match between slope angle and ball acceleration congruent with physics, but there was a systematic effect of the initial conditions: accuracy was higher when the participants started the exploration from the combination of slope and acceleration corresponding to the congruent conditions than when they started far away from the congruent conditions. In Experiment 3, participants modified the slope angle based on an adaptive staircase, but the target never coincided with the starting condition. Here we found a generally accurate performance, irrespective of the target slope. We suggest that, provided the visual scene includes sufficient cues about environmental reference and metric scale, joint processing of slope and acceleration may facilitate the detection of natural motion. Perception of rolling motion may rely on the kind of approximate, probabilistic simulations of Newtonian mechanics that have previously been called into play to explain complex inferences in rich visual scenes.

Is a matrix exponential specification suitable for the modeling of spatial correlation structures?

This paper investigates the adequacy of the matrix exponential spatial specifications (MESS) as an alternative to the widely used spatial autoregressive models (SAR). To provide as complete a picture as possible, we extend the analysis to all the main spatial models governed by matrix exponentials comparing them with their spatial autoregressive counterparts. We propose a new implementation of Bayesian parameter estimation for the MESS model with vague prior distributions, which is shown to be precise and computationally efficient. Our implementations also account for spatially lagged regressors. We further allow for location-specific heterogeneity, which we model by including spatial splines. We conclude by comparing the performances of the different model specifications in applications to a real data set and by running simulations. Both the applications and the simulations suggest that the spatial splines are a flexible and efficient way to account for spatial heterogeneities governed by unknown mechanisms.

Gaze behavior in one-handed catching and its relation with interceptive performance: what the eyes can't tell.

In ball sports, it is usually acknowledged that expert athletes track the ball more accurately than novices. However, there is also evidence that keeping the eyes on the ball is not always necessary for interception. Here we aimed at gaining new insights on the extent to which ocular pursuit performance is related to catching performance. To this end, we analyzed eye and head movements of nine subjects catching a ball projected by an actuated launching apparatus. Four different ball flight durations and two different ball arrival heights were tested and the quality of ocular pursuit was characterized by means of several timing and accuracy parameters. Catching performance differed across subjects and depended on ball flight characteristics. All subjects showed a similar sequence of eye movement events and a similar modulation of the timing of these events in relation to the characteristics of the ball trajectory. On a trial-by-trial basis there was a significant relationship only between pursuit duration and catching performance, confirming that keeping the eyes on the ball longer increases catching success probability. Ocular pursuit parameters values and their dependence on flight conditions as well as the eye and head contributions to gaze shift differed across subjects. However, the observed average individual ocular behavior and the eye-head coordination patterns were not directly related to the individual catching performance. These results suggest that several oculomotor strategies may be used to gather information on ball motion, and that factors unrelated to eye movements may underlie the observed differences in interceptive performance.

Estimating cancer incidence using a Bayesian back-calculation approach.

We propose a Bayesian hierarchical model for the calculation of incidence counts from mortality data by a convolution equation that expresses mortality through its relationship with incidence and the survival probability density. The basic idea is to use mortality data together with an estimate of the survival distribution from cancer incidence to cancer mortality to reconstruct the numbers of individuals who constitute previously incident cases that give rise to the observed pattern of cancer mortality. This model is novel because it takes into account the uncertainty from the survival distribution; thus, a Bayesian-mixture cure model for survival is introduced. Furthermore, projections are obtained starting from a Bayesian age-period-cohort model. The main advantage of the proposed approach is its consideration of the three components of the model: the convolution equation, the survival mixture cure model and the age-period-cohort projection within a directed acyclic graph model. Furthermore, the estimation are obtained through the Gibbs sampler. We applied the model to cases of women with stomach cancer using six age classes [15-45], [45-55], [55-65], [65-75], [75-85] and [85-95] and validated it by using data from the Tuscany Cancer Registry. The model proposed and the program implemented are convenient because they allow different cancer disease to be analysed because the survival time is modelled by flexible distributions that are able to describe different trends.

Cost profiles of colorectal cancer patients in Italy based on individual patterns of care.

BACKGROUND: Due to changes in cancer-related risk factors, improvements in diagnostic procedures and treatments, and the aging of the population, in most developed countries cancer accounts for an increasing proportion of health care expenditures. The analysis of cancer-related costs is a topic of several economic and epidemiological studies and represents a research area of great interest to public health planners and policy makers. In Italy studies are limited either to some specific types of expenditures or to specific groups of cancer patients. Aim of the paper is to estimate the distribution of cancer survivors and associated health care expenditures according to a disease pathway which identifies three clinically relevant phases: initial (one year following diagnosis), continuing (between initial and final) and final (one year before death). METHODS: The methodology proposed is based on the reconstruction of patterns of care at individual level by combining different data sources, surveillance data and administrative data, in areas covered by cancer registration. RESULTS: A total colorectal cancer-related expenditure of 77.8 million Euros for 18012 patients (corresponding to about 4300 Euros per capita) is estimated in 2006 in two Italian areas located in Tuscany and Veneto regions, respectively. Cost of care varies according to the care pathway: 11% of patients were in the initial phase, and consumed 34% of total expenditure; patients in the final (6%) and in the continuing (83%) phase consumed 23% and 43% of the budget, respectively. There is an association between patterns of care/costs and patients characteristics such as stage and age at diagnosis. CONCLUSIONS: This paper represents the first attempt to attribute health care expenditures in Italy to specific phases of disease, according to varying treatment approaches, surveillance strategies and management of relapses, palliative care. The association between stage at diagnosis, profile of therapies and costs supports the idea that primary prevention and early detection play an important role in a public health perspective. Results from this pilot study encourage the use of such analyses in a public health perspective, to increase understanding of patient outcomes and economic consequences of differences in policies related to cancer screening, treatment, and programs of care.

Advancing the science of cancer cost measurement: challenges and opportunities.

OBJECTIVES: Cancer accounts for a major proportion of national health expenditures, which are expected to increase in the future. This paper aims to identify major challenges with estimating cancer related costs, and discuss international comparisons, and recommendations for future research. METHODS: It starts from the experience of an international workshop aimed at comparing cancer burden evaluation methods, improving results comparability, discussing strengths and criticisms of different approaches. RESULTS: Three methodological themes necessary to inform the analysis are identified and discussed: data availability; costs definition; epidemiological measures. CONCLUSIONS: Cost evaluation is applied to cancer control interventions and is relevant for public health planners. Despite their complexity, international comparisons are fundamental to improve, generalize and extend cost evaluation to different contexts.

Modeling psychophysical data at the population-level: the generalized linear mixed model.

In psychophysics, researchers usually apply a two-level model for the analysis of the behavior of the single subject and the population. This classical model has two main disadvantages. First, the second level of the analysis discards information on trial repetitions and subject-specific variability. Second, the model does not easily allow assessing the goodness of fit. As an alternative to this classical approach, here we propose the Generalized Linear Mixed Model (GLMM). The GLMM separately estimates the variability of fixed and random effects, it has a higher statistical power, and it allows an easier assessment of the goodness of fit compared with the classical two-level model. GLMMs have been frequently used in many disciplines since the 1990s; however, they have been rarely applied in psychophysics. Furthermore, to our knowledge, the issue of estimating the point-of-subjective-equivalence (PSE) within the GLMM framework has never been addressed. Therefore the article has two purposes: It provides a brief introduction to the usage of the GLMM in psychophysics, and it evaluates two different methods to estimate the PSE and its variability within the GLMM framework. We compare the performance of the GLMM and the classical two-level model on published experimental data and simulated data. We report that the estimated values of the parameters were similar between the two models and Type I errors were below the confidence level in both models. However, the GLMM has a higher statistical power than the two-level model. Moreover, one can easily compare the fit of different GLMMs according to different criteria. In conclusion, we argue that the GLMM can be a useful method in psychophysics.

Estimation of vaccine efficacy in a repeated measures study under heterogeneity of exposure or susceptibility to infection.

Vaccine efficacy (VE) is commonly estimated through proportional hazards modelling of the time to first infection or disease, even when the event of interest can recur. These methods can result in biased estimates when VE is heterogeneous across levels of exposure and susceptibility in subjects. These two factors are important sources of unmeasured heterogeneity, since they vary within and across areas, and often cannot be individually quantified. We propose an estimator of VE per exposure that accounts for heterogeneous susceptibility and exposure for a repeated measures study with binary recurrent outcomes. The estimator requires only information about the probability distribution of environmental exposures. Through simulation studies, we compare the properties of this estimator with proportional hazards estimation under the heterogeneity of exposure. The methods are applied to a reanalysis of a malaria vaccine trial in Brazil.

Comparison between spoligotyping and IS6110 restriction fragment length polymorphisms in molecular genotyping analysis of Mycobacterium tuberculosis strains.

Spoligotyping was compared with RFLP fingerprinting analysis in the identification of Mycobacterium tuberculosis strains. Spoligotyping sensitivity was 97.6% with a specificity of 47%. The global probability for two strains clustered with spoligotyping to be clustered also with RFLP analysis was 33%; the probability for two strains clustered with RFLP analysis to be clustered also with spoligotyping analysis was 95%. However, comparing the two methods in five outbreak episodes, full concordance was evidenced between spoligotyping and RFLP. Moreover, we evaluated the presence of our 17 largest spoligotyping clusters in spoligotyping databases from Caribbean countries, London and Cuba. Only five out of 17 patterns were present in all the cohorts. The conditional probability comparing spoligotyping and RFLP methods related to these patterns resulted in very low concordance (range from 2 to 38%). In conclusion, we confirm that spoligotyping when used alone overestimates the number of recent transmission and does not represent a suitable method for wide clinical practice application. However, it allows to get a first good picture of strain identity in a new setting and in more localized or confined settings, the probability of reaching the same result compared to RFLP was 100% confirming the usefulness of spoligotyping in the management of epidemic events, especially in hospitals, prisons and close communities.