The value of tests in the diagnosis and management of glaucoma.

PURPOSE:

To assess the noneconomic value of tests used in the diagnosis and management of glaucoma, and explore the contexts and factors that determine such value.

DESIGN:

Perspective.

METHODS:

Selected articles from primary and secondary sources were reviewed and interpreted in the context of the authors' clinical and research experience, influenced by our perspectives on the tasks of reducing the global problem of irreversible blindness caused by glaucoma. The value of any test used in glaucoma is addressed by 3 questions regarding: its contexts, its kind of value, and its implicit or explicit benefits.

RESULTS:

Tonometry, slit-lamp gonioscopy, and optic disc evaluation remain the foundation of clinic-based case finding, whether in areas of more or less abundant resources. In resource-poor areas, there is urgency in identifying patients at risk for severe functional loss of vision; screening strategies have proven ineffective, and efforts are hindered by the inadequate allocation of support. In resource-abundant areas, the wider spectrum of glaucoma is addressed, with emphasis on early detection of structural changes of little functional consequence; these are increasingly the focus of new and expensive technologies whose clinical value has not been established in longitudinal and population-based studies. These contrasting realities in part reflect differences among the value ascribed, often implicitly, to the tests used in glaucoma.

CONCLUSIONS:

The value of any test is determined by 3 aspects: its context of usage; its comparative worth and to whom its benefit accrues; and how we define historically what we are testing. These multiple factors

Assessment of the impact speed and angle conditions for the EN1317 barrier tests

Roadside safety barriers designs are tested with passenger cars in Europe using standard EN1317 in which the impact angle for normal, high and very high containment level tests is 20°. In comparison to EN1317, the US standard MASH has higher impact angles for cars and pickups (25°) and different vehicle masses. Studies in Europe (RISER) and the US have shown values for the 90th percentile impact angle of 30°–34°. Thus, the limited evidence available suggests that the 20° angle applied in EN 1317 may be too low.
The first goal of this paper is to use the US NCHRP database (Project NCHRP 17–22) to assess the distribution of impact angle and collision speed in recent ROR accidents. Second, based on the findings of the statistical analysis and on analysis of impact angles and speeds in the literature, an LS-DYNA finite element analysis was carried out to evaluate the normal containment level of concrete barriers in non-standard collisions. The FE model was validated against a crash test of a portable concrete barrier carried out at the UK Transport Research Laboratory (TRL).
The accident data analysis for run-off road accidents indicates that a substantial proportion of accidents have an impact angle in excess of 20°. The baseline LS-DYNA model showed good comparison with experimental acceleration severity index (ASI) data and the parametric analysis indicates a very significant influence of impact angle on ASI. Accordingly, a review of European run-off road accidents and the configuration of EN 1317 should be performed.

Statistical Tests for Joint Analysis of Performance Measures

Recently there has been an increasing interest in the development of new methods using Pareto optimality to deal with multi-objective criteria (for example, accuracy and architectural complexity). Once one has learned a model based on their devised method, the problem is then how to compare it with the state of art. In machine learning, algorithms are typically evaluated by comparing their performance on different data sets by means of statistical tests. Unfortunately, the standard tests used for this purpose are not able to jointly consider performance measures. The aim of this paper is to resolve this issue by developing statistical procedures that are able to account for multiple competing measures at the same time. In particular, we develop two tests: a frequentist procedure based on the generalized likelihood-ratio test and a Bayesian procedure based on a multinomial-Dirichlet conjugate model. We further extend them by discovering conditional independences among measures to reduce the number of parameter of such models, as usually the number of studied cases is very reduced in such comparisons. Real data from a comparison among general purpose classifiers is used to show a practical application of our tests.