Modeling asthma evolution by a multi-state model
Boudemaghe, T.; Daurès, J.P.
Revue d'Epidemiologie et de Sante Publique 48(3): 249-255
2000
ISSN/ISBN: 0398-7620 PMID: 10891785 Document Number: 518511
Background: There are many scores for the evaluation of asthma. However, most do not take into account the evolutionary aspects of this illness. We propose a model for the clinical course of asthma by a homogeneous Markov model process based on data provided by the A.R.I.A. (Association de Recherche en Intelligence Artificielle dans le cadre de l'asthme et des maladies respiratoires). Methods: The criterion used is the activity of the illness during the month before consultation. The activity is divided into three levels: light (state 1), mild (state 2) and severe (state 3). The model allows the evaluation of the strength of transition between states. Results: We found that strong intensities were implicated towards state 2 (lambda12 and lambda32), less towards state 1 (lambda21 and lambda31), and minimum towards state 3 (lambda23). This results in an equilibrium distribution essentially divided between state 1 and 2 (44.6% and 51.0% respectively) with a small proportion in state 3 (4.4%). Conclusions: In the future, the increasing amount of available data should permit the introduction of covariables, the distinction of subgroups and the implementation of clinical studies. The interest of this model falls within the domain of the quantification of the illness as well as the representation allowed thereof, while offering a formal framework for the clinical notion of time and evolution.