The cost of the risk of work stoppage has been rising for a number of years. This increase is explained in particular by psychosocial risks. This article models the rate of prescription of work stoppages thanks to a model whose fundamentals are identical to those of the LEE CARTER model. This model takes into account three components: the year, the age group reached and the origin of the prescription of the work stoppage. Artificial intelligence algorithms will make it possible to objectivise the correlation between the components and the number of parameters to be estimated. The first part of the article consists of an analysis of disability/disability risks. We will talk about the national trends observed on this risk over the last ten years or so. In the second part of the article, we describe the model used, whose parameters will be estimated using Machine Learning algorithms and national data. The use of the results of our models in the estimation, piloting and monitoring of the risk of work stoppage (in an ORSA framework in particular and the determination of preventive actions) is studied in the third and last part.