Comparative Study of Intelligent Systems for Management of GIT Cancers

Intelligent Systems contribute in the management of different GIT cancer types. The paper discusses different types of intelligent systems, classified according to the medical task achieved, such as early detection, diagnosis and prognosis. It is found out that these types include rule-based and case-based expert systems, artificial neural networks, genetic algorithms, machine learning, in addition to data mining techniques and statistical methods. The study focuses on comparing between different techniques and tools used. The comparison results in identifying the benefits of using data mining techniques for the diagnosis task, since it is based on huge amounts of data in order to discover new patterns hence new predisposing factors. It also points out the use of expert systems in the prognosis task, since this task is mainly based on the specialist experience that should be transferred to less- experienced medical professionals. Based on the previous results, it is recommended to develop an Intelligent Tutoring System (ITS) that focuses on the early diagnosis of GIT cancers, since managing the disease depends mainly on proper diagnosis, and also to build an expert system that helps transferring GIT cancers management knowledge to medical doctors in different hospitals.


Problem Definition
The Health care forum of 2016 declared "War on Cancer" [1]. It requires more efforts to design strategies that work on decreasing cancer burden. These strategies require assembling government and industry key persons, in addition to academic professionals to work on finding measures to fight this wide spreading disease.
To deal with different types of GIT cancer management, healthcare professionals need to have sufficient data related to patient and disease details and knowledge management techniques related to new algorithms. Moreover, there is a need for intelligent systems to support different medical tasks.
Although there are several types and techniques related to management of GIT cancer, there is no clear-cut definition of the most suitable type and technique based on the required task, whether it is diagnosis, treatment or prognosis

Importance
Cancer management consumes huge amounts of funds in any country. The main cost is for treating the disease [4] Economic Impact of Cancer, American Cancer Society). Therefore, governments are working on strategies that deal with this disease in order to reduce the costs and improve the quality of life of cancer patients.

Objectives
Based on the problem definition, the study aims at achieving the following objectives: a-Discussing the current intelligent systems of the GIT cancer domain.
b-Comparing between various techniques used to decide on the suitable technique related to the medical task.

Structure of the paper
The paper will discuss the various types of intelligent systems, such as Expert Systems, Case-based learning, Intelligent Decision Support Systems, and others. Then it will discuss recent studies related to the domain, classified based on the medical task they are achieving, then make a comparison among them based on technique and tool used. Finally, results are presented and conclusions and recommendations are drawn.

Intelligent Systems
"Intelligent System (IS) can be defined as the system that incorporates intelligence into applications being handled by machines. Intelligent systems perform search and optimization along with learning capabilities. Different types of machine learning such as supervised, unsupervised and reinforcement learning can be modeled in designing intelligent systems. Intelligent systems also perform complex automated tasks which are not possible by traditional computing paradigm. Various diagnostic, robotics and engineering systems are results of intelligent procedures implemented in Intelligent System Design" "An intelligent system is a system that can imitate, automate some intelligent behaviors of human being. Expert systems, intelligent agents and knowledge-based systems are examples of intelligent systems. Currently, intelligent systems is a discipline that studies intelligent behaviors and their implementations as well as their impacts on human society". [5]

Gastrointestinal (GIT) Cancer
GIT cancer is found out to be one of the most commonly diagnosed malignancies and a major cause of mortality [6]. Some of the problems related to this major health problem worldwide are poor prognosis and limited treatment options. Hence, the main focus is on preventing this serious disease in order to reduce its incidence and mortality.
Some of the predisposing factors are diet such as eating dried fish and meat and refined carbohydrates, genetic factors, infectious diseases such as H. pylori infection, in addition to alcohol consumption and smoking [3].

Comparison from IS perspective
The following table show the different types in IS used in the field of GI cancer, classified by the three different tasks (Algorithm, Diagnosis, Prognosis) Regarding algorithms, data mining, machine learning and expert systems are used.
Therefore, although there is no single technique or tool that is used to perform a specific medical task, expert systems are mainly used for prognosis because GIT cancers follow many protocols to be treated and the prognosis relies mainly on the expertise of oncologists whereas data mining techniques contribute to the diagnosis since they mainly rely on big amounts of data in order to discover new patterns.

Conclusion and Future Work
Regarding the recommendations for future work, the following projects are suggested: