A Multi-Agent Based Modeling and Simulation Data Management and Analysis System for the Hospital Emergency Department

A Multi-Agent Based Modeling and Simulation Data Management and Analysis System for the Hospital Emergency Department

Manel Saad Saoud, Abdelhak Boubetra, Safa Attia
DOI: 10.4018/978-1-7998-2451-0.ch002
(Individual Chapters)
No Current Special Offers


In the last decades, multi-agent based modeling and simulation systems have become more increasingly used to model the dynamic and the complex healthcare systems which contain many variabilities and uncertainties such as the hospital emergency departments (ED). Modeling and creating virtual societies almost identical and similar to the reality are considered as the strongest advantages of these agents systems. However, during the dynamic development of the artificial societies, a massive volume of data, which generally contains non-express and shrouded information and even knowledge, is involved. Therefore, dealing with this data, to study and to analyze the unclear relationships and the emerging phenomena, is a well-known weakness and bottleneck that the multi-agent systems is suffering from. In conjunction, data mining techniques are the most powerful tools that can help simulation experts to tackle this issue. This paper presents an ongoing research that combines the multi-agent based modeling and simulation systems and data mining techniques to develop a decision support system to improve the operation of the emergency department.
Chapter Preview


Nowadays, the multi-agent based modeling and simulation systems have generously demonstrated their efficiency in many scientific fields, in particular with the open problems of the dynamic and the complex systems, such as the healthcare systems. Adopting simulation systems offers, in one hand, the possibility to model and to create virtual systems almost identical and similar to the real ones, where the individuals and even the organizations are directly represented with their interactions. On the other hand, using these artificial societies facilities the test and the evaluation of the possible policies and the different “what-if” scenarios, which avoid the costly and uncertain changes in the real system.

Complex systems are composed of many individuals and entities, each one of them is mainly characterized by its architecture, its behavior, and its degree of reasoning (Ferber, 1999). Based on these characteristics (Wooldridge, 2009) divided the agents into three main types: cognitive agents, reactive agents, and hybrid agents.

The cognitive agent is an intelligent agent owns a necessary knowledge base to perform its tasks and manage its interactions with the other agents and its environment. It has a symbolic representation of its environment and the agents and explicit goals and plans to decide his actions. Unlike the cognitive agent, the reactive agent has a straightforward and predefined behavior and responds only to a simple environmental stimulus. The hybrid agent represents the combination of the two previous types.

Multi-agent based modeling and simulation system of complex systems and phenomena is to model virtual societies with various individuals or agents via a computer, to observe their behaviors and to understand the relation between them.

According to (Wooldridge & R. Jennings, 1995) an agent is a software-based computer system characterized by four properties; Autonomy (Agent has the possibility to operate without a direct intervention of humans or other agents, and it has some control over his activities and inside his state), Social ability (agent connects with different agents via an agent communication language), Reactivity (Agent can perceive his environment and respond to changes that occur in it), Pro-activity (the possibility to exhibit a behavior controlled by its objectives rather than a reaction according to its environment only).

During the dynamic evolution of the virtual agents’ systems, an enormous amount of data is involved, especially, since simulation massively consume and generate the data. The data delivered generally encase underlying and covered information and even knowledge. Thus, the search in these data is of immense utility to better study and understand the operation of the system. Accordingly, a well-known weakness or bottleneck that the MAMSS is suffering from is the analysis of the unclear relationships and the phenomena that may emerge in these artificial systems. As a solution to this problem, Data Mining techniques are considered as the most powerful tools that can help simulation specialists to tackle these issues.

In conjunction, Knowledge Discovery in Database (Data Mining) (Pujari, 2001; Fayyad, Piatetsky-Shapiro, & Smyth, 1996; Hegland, 2001; Han, Kamber, & Pei, 2011) represents the process of discovering and extracting knowledge from large volumes of data. Data Mining process is divided mainly into three principal phases (Fayyad, Piatetsky-Shapiro, & Smyth, 1996): Data pre-processing, Data Mining, and Data post-processing.

Our methodology presented in this paper points out how can the multi-agent based modeling and simulation systems benefit from Data Mining techniques to solve simulation bottlenecks and to improve the study’s quality. The combination approach is demonstrated through a case study on the operation of the emergency department in the public hospital Lakhdar Bouzidi in Bordj Bou Arreridj (Algeria).

The hospital emergency department is a critical and a complex system that involves many variabilities and uncertainties. The Hospital emergency departments are often plagued with the overcrowding, the high variety of patients’ illnesses, the different resources with different skills, the uncertain patients’ arrival and the need for the same resources simultaneously. These problems contribute to long waiting time and sojourn, low quality of care, and stress situations for both patients and EDs staff.

Complete Chapter List

Search this Book: