A Bayesian Network Model for Probability Estimation

A Bayesian Network Model for Probability Estimation

Harleen Kaur (Hamdard University, India), Ritu Chauhan (Amity University, India) and Siri Krishan Wasan (Jamia Millia Islamia, India)
DOI: 10.4018/978-1-4666-5888-2.ch148
OnDemand PDF Download:
$30.00
List Price: $37.50

Chapter Preview

Top

Introduction

Deriving information from mountain of heathcare data can be too complex and voluminous to be processed by human capabilities alone. To overcome this flaw or nature of data, healthcare practioners are adopting new emergent technology to discover effective and efficient pattterns. However, medical researchers are exploiting numerous data mining techniques to find correlations or patterns among large scale databases as compared to traditional techniques for future medical diagnosis.

Data mining can be termed as an essential component for knowledge discovery process, where the vivid process determines effective and potential benefits among the data. However challenge is to establish proper exploitation strategy among health care data records which offer numerous facts in creation, dissemination and preservation of knowledge using advanced technologies, whereas, if the discovered knowledge tends to be a successful activity then gradually it can be used for futuristic decision making in healthcare organization. For instance if data of cancer patients or other diseases might consists of knowledgeable patterns which can be more expected to develop a kind of disease, so such knowledge can be used to prevent the diagnosis of patient’s disease for futuristic decision making.

Progressively Data mining and knowledge discovery are used as interdisciplinary terms to discover hidden and unknown information from large scale databases. Eventually data mining is also known as essential step in knowledge discovery process, knowledge discovery process includes several integrated preprocessing and post processing steps to discover hidden information from databases. There exists several application domain areas of data mining techniques such as medical domain for diagnosis, management survey of data, marketing area of research, statistical analysis, and geographical analysis of data and several other research areas (Arabie & Hubert, 1994; Dunham, 2003; Kaur et al., 2010; Chauhan & Kaur, 2014). Data mining techniques are highly computational techniques under certain computational circumstances to retrieve effective and efficient patterns from raw data (Fayyad et al., 1996). The output of data mining techniques can be further applied for decision making support system, to retrieve profitable environment for experts and finally provide benefits to end users. Such analyses are increasing pressure on healthcare organizations to make decision based on data mining techniques to gain insights of data. Data mining can influence medical decision making by maintaining high level of healthcare.

Medical decision making from diagnosis to patient management is becoming more and more complex due to rapid growth of knowledge during last three decades. It is possible that even with specialization and super specialization physician may not be able to make an optimal decision. Computer assisted Medical Decision making use of data mining techniques may provide a partial solution to the problem. Since medical diagnosis is probabilistic in nature, it is well suited for probabilistic formalism. Bayesian classifiers are statistical classifiers based on famous Bayes theorem of conditional probability. Thus medical diagnosis fits well into Bayesian probabilistic framework. But there are certain inherent limitations in Bayesian classification. For example, in medical applications this assumes that patient’s symptoms are independent and patient has only one disease. Moreover, exact probabilities for various attributes responsible for a disease are required. Most of the above assumptions are not true in real life, for example, a patient may have several diseases at a time. The major challenge for the applications of Bayesian networks for medical diagnosis is to represent the domain knowledge in a probabilistic formalism.

We not only need medical data of patients but we must have ability to reason. What should be there in the diagnosis? This article examines Bayesian classification technique for exploration of medical diagnosis for various attributes resulting in development of futuristic decision making. The proper exploitation of Bayesian strategy can be deployed for further investigation and refinement of medical diagnosis. In this paper we have utilized Bayesian classification technique for patients admitted to hospital for Right Heart Catheterization and used probability technique to evaluate different characteristics or features with sustainability rate among the patients.

Key Terms in this Chapter

Artificial Intelligence: The science that deals with machine performance tasks that require intelligence based on humans.

Bayesian Techniques: The technique of classifying data items in a data bank.

Genetic Algorithm: A search algorithm to enable you to locate optimal binary strings by processing an initial random population of binary strings by performing operations.

Bayesian Classifiers: A class of learning algorithm to find optimal classification using probabilistic theory.

Machine Learning: The process of generating a computer system that is capable of independently acquiring data and integrating that data to generate useful knowledge.

Neural Networks: A class of machine learning algorithm that works on the structure of biological nervous systems. A class of machine learning algorithm consisting of multiple nodes that communicate through their connecting synapses.

Complete Chapter List

Search this Book:
Reset