Multi-Agent System Based on Data Mining Algorithms to Detect Breast Cancer

Introduction

We live in the data evolution. Since the data come from a wide range of sources, including social networking sites, supply chains and databases; it is usually unstructured in the absence of a particular format or layout. Therefore, the researchers must process the incoming data for useful information. Data mining is the process in which intelligent methods can extract interesting data models and knowledge from large amounts of data (Rao, 2017). However, the speed at which data is produced is very high, and they need efficient methods of operating to keep up to date. Distributed Data Mining (DMD) aims at extracting a useful model from distributed heterogeneous databases in order, for example, to compose them within a distributed knowledge base and use it for decision-making purposes. DMD can also be useful in environments with multiple computing nodes connected over high speed networks. Although data can be quickly centralized using the relatively fast network, proper compute load balancing between a group of nodes may require a distributed approach. The distributed nature of agent extraction brings several benefits to data mining, such as scalability, scalability, reliability, security, interactivity, and high speed (Bellifemine, Caire, Poggiet, & Rimassa, 2008). Agents can be used to automate various tasks such as data selection, data cleansing, and data preprocessing for classification and representation of knowledge. As an emerging field, a lot of research can be done in this area. for this chapter authors have found that, moreover, it is difficult for a specialist to diagnose in a patient whether the patient has breast cancer or not, to confirm his presence or to determine his characteristics (his extension, his aggressiveness, ....). The current situation has motivated research in this field and the need to automate medical diagnosis has become indispensable. The aim of this system is to give more autonomy and initiative to the various software modules specialized in medical diagnosis and which can interact for share their knowledge as human experts. Breast cancer remains the first female cancer in the world, it is a real social problem. The frequency of breast cancer varies greatly from country to country and the factors involved are multiple: genetic factors, role of diet. The issue of breast cancer detection leads researchers, specialists in the field to look at other trends, new technologies other than human to address this real social problem. The goal of this chapter is to create a new approach that will help determine whether a patient has benign or malignant cancer following multiple descriptors. To achieve this goal, researchers propose a solution based on the concept of multi-agent systems. they perform a coupling between two major areas of Medicine and Computer Science for the diagnosis of breast cancer and the agent paradigm respectively. The purpose of this coupling is to improve the performance and efficiency of medical diagnostic systems. The peculiarity of their the approach is the development and definition of a model based on an architecture composed of distinct parts, each part will diagnose whether the cancer is malignant or benign from a database in a specific way, but able to communicate to share their knowledge. The multi agent paradigm in their approach would be applied in several contexts: Each agent in this system will make its own diagnosis for the same learning base (autonomy). The final decision will be made after a majority vote: the patient will be assigned to the class that has been appointed by the large number of agents.