RESUMEN
Translational research requires large patient populations. A single research institute is not able to build up such a population in a short period of time. The String of Pearls Initiative (in Dutch "Parelsnoer Initiatief", PSI) is a joint effort by the eight academic medical centers in the Netherlands to built an infrastructure for joint biobanking as to meet this challenge of establishing large collections of data and samples in relevant medical domains.
RESUMEN
In 2003 a workshop was held in Innsbruck, Austria, on the topic of evaluation of ICT applications in Health Care. A result of that workshop was the "Declaration of Innsbruck". In the current paper we will further elaborate on this declaration and discuss some of the activities that are currently undertaken as supportive measures to enable the realization of the vision expressed in the declaration.
Asunto(s)
Atención a la Salud/organización & administración , Gestión de la Información , Austria , EducaciónRESUMEN
OBJECTIVE: An automated feedback system that produces comments about the non-adherence of general practitioners (GPs) to accepted practice guidelines for ordering diagnostic tests was developed. Before implementing the automated feedback system in daily practice, we assessed the potential effect of the system on the test ordering behaviour of GPs. DESIGN: We used a randomised controlled trial with balanced block design. SETTING: Five times six participant groups of GPs in a computer laboratory setting. INTERVENTION: The GPs reviewed a random sample of 30 request forms they filled in earlier that year. If deemed necessary, they could make changes in the tests requested. Next, the system displayed critical comments about their non-adherence to the guidelines as apparent from the (updated) request forms. SUBJECTS: Twenty-four randomly selected GPs participated. MAIN OUTCOME MEASURES: The number of requested diagnostic tests (17% with 95% confidence interval [CI]: 12-22%) and the fraction of tests ordered that were not in accordance with the practice guidelines (39% with 95% CI: 28-51%) decreased due to the comments of the automated feedback system. The GPs accepted 362 (50%) of the 729 reminders. IMPLICATIONS: Although our experiment cannot predict the size of the actual effect of the automated feedback system in daily practice, the observed effect may be seen as the maximum achievable.
Asunto(s)
Sistemas de Apoyo a Decisiones Clínicas , Medicina Familiar y Comunitaria/organización & administración , Guías de Práctica Clínica como Asunto , Atención Primaria de Salud/métodos , Sistemas Recordatorios , Femenino , Humanos , Masculino , Persona de Mediana EdadRESUMEN
In this paper we establish a mathematical framework in which we develop measures for determining the contribution of individual features to the performance of a classifier. Corresponding to these measures, we design metrics that allow estimation of the importance of features for a specific multi-layer perceptron neural network. It is shown that all measures constitute lower bounds for the correctness that can be obtained when the feature under study is excluded and the classifier rebuilt. We also present a method for pruning input nodes from the network such that most of the knowledge encoded in its weights is retained. The proposed metrics and the pruning method are validated with a number of experiments with artificial classification tasks. The experiments indicate that the metric called replaceability results in the tightest error bounds. Both this metric and the metric called expected influence result in good rankings of the features.