الفهرس 
About the LiverOnly 14 pages are availabe for public view
 |  | from | 165 |  |  |
| | | from | 165 | | |
Abstract
Statistical Model to Determine the Factors
that Affect Liver Cancer
. Introduction
Liver cancer is one of the most malignant tumors with a high mortality
rate, aggressive growth behavior and a high recurrence rate. Hepatocellular
carcinoma (HCC) which is the most frequent type of cancer liver is a major
worldwide public health concern. Worldwide, hepatocellular carcinoma
ranks sixth among cancer incidence, and is the second leading cause of
cancer death, with an estimated , cases of death in the world during
. Hepatocellular carcinoma (HCC) is a major public health problem in
Egypt and its incidence is increasing. It also has a rising incidence in Egypt
mostly due to high prevalence of viral hepatitis and its complications.
This study aims to identify the risk factors that affect the survival of
hepatocellular carcinoma (HCC) patients, using the statistical models to
predict survival of HCC patients, especially because of spread of viral
hepatitis C infection in Egypt, with its impact on the probability of a decline
in survival of these patients. It changes fast and can stay inside the patient‘s
body for many years causing great damage and complications to liver, and
causing a problem to human health. Hepatitis C virus causes many serious
diseases which are life-threatening leading to death. Hepatitis C virus still
concerns and puzzles human minds. Despite some hepatitis viruses cause acute liver inflammation leading to liver dysfunction, but hepatitis C virus
remains the most virulent deadly virus of the liver. Some of the risk factors
such as patient‘s gender, patient‘s age group, hepatitis C virus, child score,
alpha fetoprotein (AFP), and performance status (PS) have been studied in
this study using discriminant analysis model, logistic regression model,
artificial neural networks model, and classification and regression tree model
for prediction of survival of HCC patients.
. Nature of the problem
Hepatocellular carcinoma has a rising incidence in Egypt. The problem of
increasing incidence of hepatocellular carcinoma rate accompanying the
increase in death rate due to this disease, has needed to identify the risk
factors and causes, and to develop assumptions that may affect the decrease
in survival of HCC patients.
Egypt has the highest hepatitis C virus (HCV) prevalence worldwide,
which accompanies great dangers to the liver. Due to the increasing
incidence of chronic viral hepatitis C, there is a state of uncertainty in
answering the question: Is hepatitis C virus a dangerous and major factor that
decreases the survival of HCC patients or not? Also, if it‘s one of the major
factors, is it accompanied by other factors?
. Objectives of the Study
The early detection and evaluation of risk factors which might affect the
decrease of survival rate of hepatocellular carcinoma (HCC) patients is very
important. The prediction of risk factors is an important pivot in saving the
lives of these patients, especially HCV patients. Also, it may help doctors to
focus on these factors and inform patients to avoid it. The usage of statistical methods to identify risk factors would help to improve the survival of HCC
patients.
This study aims to identify the independent variables that affect the
survival of hepatocellular carcinoma patients‘ group membership, and the
use of statistical models to predict and explain the relationship between the
studied covariates and survival of hepatocellular carcinoma patients. Also, it
proposes a statistical classification model to determine group membership.
. Source of Data and Variables of the Study
from ( ) registered patients at Ain Shams university hospitals,
radiation oncology & nuclear medicine department, gastroenterology unit,
Cairo, Egypt. The data contains different types of gastrointestinal cancer
patients covering a period of years from year to year . Only
( ) patients meet the study assumptions as follows:
a) Hepatocellular carcinoma patients with hepatitis C virus infection.
b) Hepatocellular carcinoma patients without hepatitis C virus infection.
Variables of the Study:
The dependent variable: Survival of HCC patients.
The independent variables:
. Patient‘s gender.
. Patient‘s age group.
. Hepatitis C virus (HCV).
. Child Score.
. Alpha Fetoprotein (AFP).
. Performance Status (PS). . Results of the Study
The discriminant analysis model, and the logistic regression model,
showed that hepatitis C virus, age groups, child score are the risk factors that
affect the survival of HCC patients.
While the artificial neural networks model focused on multilayer
perceptron network algorithms classification and regression tree model,
showed that hepatitis C virus, age groups, child score, and alpha fetoprotein
are the risk factors that affect the survival of HCC patients.
The correct classification percentage for the discriminant analysis model
is %, and the hit ratio for the binary logistic regression model is . .
Also, the hit ratios for the artificial neural networks model and classification
and regression tree model are % and . % respectively.
This means that, the artificial neural networks model has high
classification accuracy and it is fit for prediction, so depending on the ANNs
model for prediction of the survival of HCC patients.
. The Outline of the Study
This study is presented in five chapters summarized as follows:
Chapter One: Introduction
This chapter starts with an overview on liver cancer focusing on its types,
and its causes. It also provides an overview on hepatitis C virus, focusing on
its types, symptoms, its spread in Egypt and its relation to HCC. This chapter
then illustrates the importance and the objectives of the study.Chapter Two: Discriminant Analysis and Logistic Regression Analysis
This chapter presents the discriminant analysis function. Also, it presents
the logistic regression analysis, logistic regression curve and logit function.
Chapter Three: Artificial Neural Networks and Classification and
Regression Tree
This chapter presents the theoretical background of artificial neural
networks analysis focusing on multilayer perceptron network algorithms.
Then the chapter introduces the classification and regression tree analysis,
shape of the tree and stopping criteria.
Chapter Four: Statistical Application Techniques
This chapter presents the applications for discriminant analysis, logistic
regression analysis, artificial neural network analysis focusing on multilayer
perceptron network algorithms, and classification and regression tree
analysis. Then the chapter introduces the results and summarized comparison
between the four techniques considered using SPSS v . , STISTICA v . ,
and DTREG v . softwares.
Chapter Five: Conclusions, and Recommendations
Keywords
Primary Liver Cancer, Hepatocellular Carcinoma, Hepatitis C Virus,
Discriminant Analysis, Binary Logistic Regression, Artificial Neural
Networks, Classification and Regression Tree.