A hybrid diagnosis model for determining the types of the liver disease
Introduction
Liver is the largest internal organ in the human body, playing a major role in metabolism and serving several vital functions like decomposition of red blood cells. Liver disease usually caused by inflammation or damaged hepatocytes, registers a tenacious presence on the list of top ten fatal diseases in Taiwan. On the worldwide scale, live cancer, or more specifically hepatocellular carcinoma (HCC), remains the third most common cause of cancer-related deaths and the fifth-most frequent cancer with an estimated 560,000 new cases every year [1]. Symptoms of liver diseases tend to be negligible in the initial stage, and the condition is usually quite serious when the symptoms finally become obvious enough to attract attention. Early, effective detection of liver diseases are thus an issue of paramount importance.
Medical diagnosis relies to a great extent on a physician’s practical experiences. Yet it may take years for a physician, especially a new or junior one, to accumulate sufficient experiences. To help physicians tackle with diagnostic problems, various disease diagnosis models based on statistics and classification have been developed as they treat medical diagnosis as a decision making process during which the physician induces the diagnosis of a new and unknown case from his or her clinical experience.
Most previous researches on the development of disease diagnosis models use statistical methods for modeling. Such approaches, however, require assumptions and are usually adopted to analyze linear data. They are thus less capable of handling massive and complicated nonlinear and dependent data [2], [3]. Recent studies have accordingly adopted potentially more effectual alternatives like genetic algorithms (GA), artificial neural networks (ANN), rough sets, and support vector machines (SVM). However, most of these studies tend to concentrate on identifying the presence of a certain disease in a patient. Not many diagnosis models have so far been developed to move beyond the detection of a disease to suggest the nature, i.e. the type, of the disease.
This study accordingly introduces an intelligent liver diagnosis model (ILDM) using ANN to distinguish between healthy and diseased liver. Furthermore, analytic hierarchy process (AHP) is integrated with case-based reasoning (CBR) to diagnose the types of liver disease in patients. The constructed model can be expected to help physicians achieve more accurate and comprehensive liver diagnosis.
The structure of this paper is organized as follows. Section 2 presents a comprehensive literature review on liver disease diagnosis with an introduction to the ANN and CBR techniques. Section 3 outlines the architecture of the proposed ILDM. Section 4 details the experiments and results of liver disease diagnosis, and Section 5 summarizes the paper with brief conclusions and suggestions for future researches.
Section snippets
Artificial neural networks (ANN)
ANN, originally derived from neurobiological models, is a system composed of highly interconnected and interacting processing units. The most important feature of neural networks is their ability to learn. Just like human brain, neural networks can learn by example and dynamically modify themselves to fit the data presented. Therefore, ANN has attracted many researchers and emerged as one of the most popular tools for pattern recognition and classification [4]. Classification rules are useful
The proposed ILDM (intelligent liver diagnosis model)
Disease diagnosis mainly relies on the physician’s clinical experience and relevant examinations. In order to simplify the diagnosis and enable the physician to make a judgment about the existence of liver disease, the study develops an ILDM that incorporates two processing phases: an initial, classifying phase for examining the existence of a liver disease and a subsequent, concluding phase for diagnosing the type of the existing liver disease. In the classifying phase, ANN is adopted to
Experiment and result
Essential health examination data of 510 outpatient visitors to a medical center in Taiwan during the 1 year from March 2005 to February 2006 was collected as the cases in the dataset. With the assistance of doctors and medical specialists of liver conditions, 210 cases were found to be healthy, and 300 cases were diagnosed with liver conditions (chronic hepatitis, alcohol hepatitis, liver cirrhosis, B hepatitis, and others). The health examination data incorporated the results of an examinee’s
Conclusion
In spite of the constant advancement in the field of medical sciences, diagnosis of diseases remains a challenging task. Liver disease in particular is not easily discovered at its initial stage; early diagnosis of this leading cause of mortality is therefore highly important. As a part of the ongoing efforts to make diagnosis more effective and efficient, this study accordingly develops a two-phase intelligent diagnosis model aiming to provide a comprehensive analytic framework to raise the
Conflict of interest statement
None declared.
Acknowledgements
We would like to thank Chun-Ming Huang for collecting questionnaire data and providing clinical attributes’ weights. We are grateful to the five physicians who participated in the survey. We also appreciate the two anonymous reviewers for their valuable comments.
Chun-Ling Chuang is an assistant professor in the Department of Information Management at Kainan University, Taiwan. She received her M.S. and Ph.D. degrees in Department of Industrial Engineering from the University of Iowa. Her present research interests include applications of hybrid soft computing models in data mining, optimal solution, and economical models for decision making, knowledge management and classification.
References (28)
- et al.
A hybrid approach for feature subset selection using neural networks and ant colony optimization
Expert Systems with Applications
(2007) - et al.
Utilization of artificial neural networks in the diagnosis of optic nerve diseases
Computers in Biology and Medicine
(2006) - et al.
Predictors of urinary tract infection based on artificial neural networks and genetic algorithms
International Journal of Medical Informatics
(2007) - et al.
Mining the breast cancer pattern using artificial neural networks and multivariate adaptive regression splines
Expert Systems with Applications
(2004) - et al.
Case-based learning of plans and goal states in medical diagnosis
Artificial Intelligence in Medicine
(1997) - et al.
Quantitative study of ductal breast cancer patient targeted prognosis: an exploration of case-base reasoning
Pathology, Research and Practice
(1997) - et al.
Clinical decision-support systems for intensive care units using case-based reasoning
Medical Engineering and Physics
(2000) - et al.
A new hybrid case-based architecture for medical diagnosis
Information Sciences
(2004) Using case-based reasoning to diagnostic screening of children with developmental delay
Expert Systems with Applications
(2005)- et al.
Case-based reasoning for medical knowledge-based systems
International Journal of Medical Informatics
(2001)
An intelligent model for liver disease diagnosis
Artificial Intelligence in Medicine
Designing a neural network for forecasting financial and economic time series
Neurocomputing
Global cancer statistics 2002
CA: A Cancer Journal for Clinicians
Fatty liver in non-alcoholic non-overweight Japanese adults: incidence and clinical characteristics
Journal of Journal of Gastroenterology and Hepatology
Cited by (60)
Multimodal biomedical image fusion method via rolling guidance filter and deep convolutional neural networks
2021, OptikCitation Excerpt :Compared with the single modal image, the fusion image has more information and better visual quality. Therefore, image fusion algorithms have been widely used in different fields, such as computer aided detection and diagnosis [1–4], image retrieval and classification [5–8] and object recognition [9–11]. Although there are extensive fusion algorithms proposed, most of them are based on the multiscale transformation [12–14] (MST).
Liver disease screening based on densely connected deep neural networks
2020, Neural NetworksCitation Excerpt :This work is based on LFT and demographic information. Lin and Chuang extend this work by changing CART into Artificial Neural Networks (ANN), and adding health and lifestyle survey results as features (Lin & Chuang, 2010). Zhou et al. combined support vector data description (SVDD) with data visualization techniques and the glowworm swarm optimization (GSO) algorithm to improve liver disease diagnostic accuracy (Zhou, Zhang, Shi, Shi, & Zheng, 2014).
Liver Cancer Classification Using Single Pass Neural Networks Based on Ultrasound Images: A Review
2023, Wireless Personal CommunicationsA survey on data mining and machine learning techniques for diagnosing hepatitis disease
2023, International Journal of Biomedical Engineering and TechnologyLiver Disease Detection: A Review of Machine Learning Algorithms and Scope of Optimization
2022, International Journal of Healthcare Information Systems and InformaticsA Hybrid Approach for Prediction and Stage Wise Classification of Liver Failure
2022, 7th International Conference on Communication and Electronics Systems, ICCES 2022 - Proceedings
Chun-Ling Chuang is an assistant professor in the Department of Information Management at Kainan University, Taiwan. She received her M.S. and Ph.D. degrees in Department of Industrial Engineering from the University of Iowa. Her present research interests include applications of hybrid soft computing models in data mining, optimal solution, and economical models for decision making, knowledge management and classification.
Rong-Ho Lin is an associate professor in the Department of Industrial Engineering and Management at National Taipei University of Technology. He received the Bachelor degree in Chun-Yuan Christian University, Taiwan, the M.S. degree in Department of Mathematics and Computer Science from Minnesota State University, and the Ph.D. degree in the Industrial Management and Engineering from the University of Iowa. He has worked in various enterprises for 10 years and his research activities include applications of artificial intelligence and hybrid soft computing models in decision making and supply chain management.