Statistical Agent Based Modelization of the Phenomenon of Drug Abuse

The project consists of the development of an agent-based model for the phenomenon of drug abuse.

It is an interdisciplinary piece of work in which the concepts and methods of statistical physics are applied to investigate the socially relevant phenomenon of drug abuse. Until 1990 theoretical statistical physics consisted of the study of fundamental models like Ising problems and variations, percolation models, spin glass models, chaos models, fractal growth models, and self-organized-criticality models, just to mentions the main ones. The most challenging problems in the field were the understanding of the proprieties of these difficult models with the idea that this would form the basis of the field.

Since that period, there have not been many new basic models introduced. Statistical Physics and Complexity have then evolved toward the application to specific problems, many of them in interdisciplinary areas. The explosion of the complex network analysis has shown that many phenomena, even those not closely related to the original physical area, can be represented with these tools.

The importance of this and other applications now have to be evaluated with respect to the impact these ideas and methods have in other fields where there are applied. This situation poses new challenges and problems that have been discussed in various articles and editorials. My interest in the phenomenon of drug abuse was triggered by the fact that Prof. Pietronero and I were invited to be part of a commission of the Italian government to monitor and control this problem. Most of the people on the committee (Prevo.Lab) were associated with the police force or were part of the medical field. However, a few of us were from apparently unrelated areas, like economics, marketing, and, in my case, statistical physics. Being part of this committee has exposed me to an exorbitant amount of information about the problem from the best national experts. We then decided to take this unique opportunity to construct a model for the phenomenon of drug abuse that could provide a scientific framework for its analysis.

The model we developed is part of the class of the Agent Based Models (ABMs), which have recently been used in many socio-economic areas. The relation of ABM models with statistical physics and in social behaviors is clear because one attempts to describe the competition between interaction and noise, the heterogeneity of the agents, the origin of large fluctuations, and the spontaneous development of critical situations. The available literature presents only highly simplified and preliminary approaches in this direction and we believe our model provides a basis for a scientific approach to this problem, its understating and possibly its optimal control.

The model is directly related to the concepts and the parameters used by professionals in the field at the international level and it makes optimal use of all the available information. This allows for a direct comparison of the present and future data and can be easily generalized to explain a more complex and realistic environment. At the present level, the model permits to identify the crucial, most important parameters, and as well as those that play a minor role. This leads to a basic understanding of the complexity of the phenomenon and to the possibility of analyzing how the situation changes thanks to some social perturbations.


Di Clemente, R. & Pietronero Statistical Agent Based Modelization of the Phenomenon of Drug Abuse, Scientific Reports 2, 532 DOI:10.1038/srep00532 (2012)
Some Italian press agency and Newspapers published a news on the relevance of my paper for studying drugs Abuse phenomena. Here the collection to the main agencies and news (Italian).