A principal components analysis self-organizing map.

We propose a new self-organizing neural model that performs principal components analysis. It is also related to the adaptive subspace self-organizing map (ASSOM) network, but its training equations are simpler. Experimental results are reported, which show that the new model has better performance than the ASSOM network.     

The diminished efficiency and altered dynamics of respiration in experimental pulmonary congestion

Quantitative measurements of the relative efficiency of respiration in the normal lung, as contrasted with the congested lung, have not been made in the intact experimental animal. In 1934 Christie and Meakins,¹ utilizing simultaneous determinations of dynamic intrapleural pressure and of tidal exchange in patients with congestive heart failure, demonstrated a marked decrease in pulmonary distensibility. With appropriate treatment, distensibility was found to increase as the patient improved. In the present study, an approach similar to that of these authors has been used, and studies of the efficiency of respiration have been made on normal dogs at rest and during hyperpnea (produced by the inhalation of carbon dioxide) and contrasted with the relative efficiency of respiration after pulmonary congestion induced by rapid venous infusion. With this approach, a quantitative comparison of the work necessary to achieve a given amount of tidal exchange has been possible in an animal with normal lungs and in the same animal after severe pulmonary congestion. From measurements of tidal exchange, ventilation, and intrapleural pressure fluctuations, a representation of the dynamic changes in these factors during progressive pulmonary congestion has been obtained.     

L’imputation multiple des données manquantes aléatoirement : concepts généraux et présentation d’une méthode Monte-Carlo

BackgroundStatistical analysis of a data set with missing data is a frequent problem to deal with in epidemiology. Methods are available to manage incomplete observations, avoiding biased estimates and improving their precision, compared to more traditional methods, such as the analysis of the sub-sample of complete observations.MethodsOne of these approaches is multiple imputation, which consists in imputing successively several values for each missing data item. Several completed data sets having the same distribution characteristics as the observed data (variability and correlations) are thus generated. Standard analyses are done separately on each completed dataset then combined to obtain a global result. In this paper, we discuss the various assumptions made on the origin of missing data (at random or not), and we present in a pragmatic way the process of multiple imputation. A recent method, Multiple Imputation by Chained Equations (MICE), based on a Monte-Carlo Markov Chain algorithm under missing at random data (MAR) hypothesis, is described. An illustrative example of the MICE method is detailed for the analysis of the relation between a dichotomous variable and two covariates presenting MAR data with no particular structure, through multivariate logistic regression.ResultsCompared with the original dataset without missing data, the results show a substantial improvement of the regression coefficient estimates with the MICE method, relatively to those obtained on the dataset with complete observations.ConclusionThis method does not require any direct assumption on joint distribution of the variables and it is presently implemented in standard statistical software (Splus, Stata). It can be used for multiple imputation of missing data of several variables with no particular structure.     

Lien précarité – durée et complexité des séjours hospitaliers en secteur de court séjour

BackgroundSeveral studies have shown that socioeconomic deprivation is associated with increased hospitalization lengths of stay (LOS) and costs. Yet, the French DRG-based information system (PMSI) does not take deprived situations into account. Hence, we aimed at extracting routinely available variables measuring deprivation from the Hospital Information System and at assessing their association with severity of illness and hospital LOS.MethodsWe performed record linkage between the PMSI database concerning stays of patients aged more than 16 years in the short-stay sector of Assistance publique–Hôpitaux de Paris in 2007 and an administrative database which provided the following deprivation measures: recipients of Couverture Médicale Universelle (basic or complementary health insurances adapted for underprivileged French citizens) or Aide Médicale d’État (health and medical emergency insurances adapted for underprivileged non French citizens living in France) and homeless patients. We compared length of stays showing a deprivation measure to others after adjustment on morbidity, age and sex.ResultsAmong 352,721 stays, the prevalence of the deprivation measures ranged from 0.71% for “homelessness” to 6.24% for complementary Couverture Médicale Universelle. Stays showing a deprivation measure had specific illnesses and had more frequently associated comorbidities or complications than others. After adjustment, deprivation measures were associated with significantly increased LOS (by 5% for Couverture Médicale Universelle to 48% for emergency Aide Médicale d’État.ConclusionRoutine extraction of deprivation measures from Hospital Information Systems is feasible. Age, sex and illness being equal, these deprivation measures were associated with more complicated cases and increased LOS. We recommend that case mix-based hospital prospective payment systems take socioeconomic deprivation into account.