International Journal of Computational Intelligence Research & Applications, 4(2), July-December 2010, pp. 181-186 © Serials Publications, New Delhi (India)
Machine Learning Based Orthopedic Expert Advisory System Using Decision Tree and Back Propagation Algorithms N. V. Ramana Murty1*, M. S. Prasad Babu2, S. V. N. L. Narayana3 & I. S. Siva Rao4
Abstract: Machine Learning is new generation Computing Technology and Capability. An Expert System, a typical AI application, has a large database and allows the user to interact and simulates the knowledge and expertise of a human expert. A medical expert system takes the patient details as input and consults the knowledge base and finally displays the diagnosis, recommended treatment and health advice. This paper ‘Machine Learning Based Orthopedic Expert Advisory System using Decision Tree and Back Propagation Algorithms’ is a Rule based, neural network Based and Decision Tree Based Medical expert system developed using the techniques of backward chaining mechanism and Back-Propagation Algorithm by collecting the expert knowledge in the form of rules and facts, from a domain expert. Orthopedic expert advisory system using Machine Learning intelligent techniques is a web based java application aimed at providing information, health advice to the rural people at their doorsteps. This Expert System can be implemented in all rural languages, as it can be easily understandable by everyone all over the world. Keywords: Artificial Intelligence- Rule Based Expert System-J2EE Environment-Rural Based Health Advice-Orthopedics.
1. INTRODUCTION
Expert Advisory System providing access to the rural Orthopedic surgery or orthopedics is the branch of surgery people at their footsteps. Orthopedic information system concerned with acute, chronic, traumatic, and overuse contains static pages in html providing information about injuries and other disorders of the musculoskeletal system. different Common Diseases, Common Symptoms, Orthopedic surgeons address most musculoskeletal ailments Investigations, Drugs, Services and Preventive Measures including arthritis, trauma and congenital deformities using collected from the domain expert in the field of Orthopedic. both surgical and non-surgical means. Many orthopedic The Expert Advisory System contains Orthopedic surgeons elect to do further subspecialty training in Knowledge Base and provides JSP based dynamic pages, programs known as ‘fellowships’ after completing their which contains simulated expert advice on the subject, to residency training. Fellowship training in an orthopedic the end users when they interact with the expert system on subspecialty is typically one year in duration (sometimes line and submit answers to queries. two) and usually has a research component involved with The basic components of the system are: the clinical and operative training. Examples of orthopedic subspecialty training in the US are: Hand surgery (also Expert: performed by Plastic Surgeons), Shoulder and elbow surgery, Total joint reconstruction (arthroplasty), Pediatric • Experts are Orthopedicians and researchers in the field of Orthopedics orthopedics, Foot and ankle surgery, Spine surgery, Musculoskeletal oncology, Surgical sports medicine, Knowledge Engineer: Orthopedic trauma. This is a java application consisting of two modules • The role of the Knowledge engineer is to collect expert advice from several experts and to convert them into namely Orthopedic Information System and Orthopedic rules and facts. 1
M.Tech, (Ph.D), Associate Professor, Department of Computer Applications, G.V.P. Degree College (A),Visakhapatnam.
2
Professor, Department of CS&SE, A.U.College of Engg, Andhra University, Visakhapatnam.
3
Senior Assistant Professor, M.R.P.G. College, Vizianagaram.
4
Associate Professor, Raghu College of Engineering, Vizianagaram.
*
Correspondence Author:
[email protected]
182
•
International Journal of Computational Intelligence Research & Applications
The Knowledge Engineer is responsible for update the changes made by the experts according to their research findings.
Store: The information collected through experts is stored as a database (Knowledge Base) that serves as a repository for quick processing and future retrieval. The system stores the information in html files.
Knowledge Base:
1) About Orthopedic
2)
Symptoms
•
Knowledge Base consists of facts and rules as text files, images and video clips about orthopedic field.
3) Diseases
4)
Preventions
•
User may be a patient/ doctors/rural people.
User Interface:
•
The user interface is to exchange information between the user and the inference engine, which involves graphical user controls such as Buttons, Labels, List Boxes, etc are used to make the system user friendly.
Inference Engine:
•
The inference engine design for information system uses forward chaining strategy and the expert advisory system uses backward chaining. The main objective of the inference engine is inferring the data from the knowledge base by using different rules of AI like modes ponens, resolution methods and interact with the problem solving unit that is nothing but front end of the paper.
5) Common Drugs The System Stores the information related to expert design in knowledge base in the following ways. Rules: A set of rules, which constitute the program, stored in a rule memory of production memory and on an inference engine required to execute the rules. Dataset: The monitoring data is in the MySQL database. It can be used as any other data stored in a database. This greatly increases the opportunity with which you can conduct post-analysis of the monitoring data. 4. MACHINE LEARNING ARCHITECTURE OF ORTHOPEDIC EXPERTADVISORY SYSTEM SYMPTOMS
ID3 ALGORITHM
2. PROPOSED SYSTEM The proposed system is divided into two aspects: 1) Information System
Validates by Domain expert
2) Advisory System In Information system, the user can get all the static information about different diseases, Symptoms of the Diseases of Orthopedic related. In Advisory System, the user is having an interaction with the expert system online; the user has to answer the questions asked by the Expert System. Depends on the response by the user the expert system decides the disease and displays its control measure of disease. 3. FUNCTIONAL REQUIREMENTS
(System 1) RULED BASED SYSTEM
DISPLAY ADVICE TO END USER
(System 2) OPTIMIZATION ALGORITHM
Display Probable Disease
KNOWLEDGE BASE
SYMPTOMS
ID3 ALGORITHM
Inputs: The system needs the information about the symptoms from the user to produce the output. Outputs: The outputs of the system will be:
Validates by Domain Expert
1) Information Diseases 2) Small Description about the disease 3) Preventions KNOWLEDGE BASE
Machine Learning Based Orthopedic Expert Advisory System Using Decision Tree and Back…
Architecture of Subsystem–II (RULED BASED SYSTEM): (System 1) RULED BASED SYSTEM
DISPLAY ADVICE TO END USER
183
Else below this new branch add the subtree ID3 (Examples(vi), Target_Attribute, Attributes – {A})
•
End
•
Return Root
Rule Based System (System-1)
KNOWLEDGE BASE
5. DECISION TREE AND ID3 ALGORITHM In decision tree learning, ID3 (Iterative Dichotomiser 3) is an algorithm used to generate a decision tree invented by Ross Quinlan.
In the Rule Based System the System takes the Symptoms as Input and produce the Exact Disease with all the facts and Rules that matches with in the Knowledge base. This Rule Based System Consists of Knowledge Base, Inference Engine, User Interface, Expert and the User. Algorithm:
Repeat
•
Collect the rules whose conditions match facts in Working Memory.
•
If more than one rule matches
The ID3 algorithm can be summarized as follows: 1. Take all unused attributes and count their entropy concerning test samples
o
2. Choose attribute for which entropy is maximum
•
3. Make node containing that attribute The algorithm is as follows:
ID3 (Examples, Target_Attribute, Attributes) •
Create a root node for the tree
•
If all examples are positive, Return the single-node tree Root, with label = +.
•
If all examples are negative, Return the single-node tree Root, with label = –.
•
If number of predicting attributes is empty, then Return the single node tree Root, with label = most common value of the target attribute in the examples.
•
Otherwise Begin
o
A = The Attribute that best classifies examples.
o
Decision Tree attribute for Root = A.
o
For each possible value, vi, of A,
Add a new tree branch below Root, corresponding to the test A = vi. Let Examples (vi), be the subset of examples that have the value vi for A If Examples (vi) is empty Then below this new branch add a leaf node with label = most common target value in the examples
Use conflict resolution strategy to eliminate all but one
Do actions indicated by the rules (add facts to WM or delete facts from WM)
Until problem is solved or no condition match
If the system1 (Rule Based System) unable to produce the exact disease then the system2 (Optimization Algorithm) explained below starts performing its work. Optimization Algorithm – Back Propagation Algorithm (System–2): The Base idea of this algorithm is taken from the concept of Neural Networks auto associative memory model. The sets of diseases and the symptoms are mapped onto a neural network and they will be trained using back propagation Algorithm. *0 indicates that the symptom is not present for that particular disease and 1 indicates that the symptom is present for that particular disease. The user is triggers queries such a whether symptoms1 to symptoms6 are present are not and they are stored in symptom vector as [1,0,1,1,1,0,1,1] indicating 1 for symptom present and 0 for symptom not present. Once the Symptom vector is submitted to the system (inference Engine) it compares the symptom vector with the matrix memory Result vector, which consists of value less than or equal to number of symptoms in the memory matrix.
184
International Journal of Computational Intelligence Research & Applications
Figure 1: Neural Network Based Algorithm
6. RESULTS & DISCUSSIONS
Machine Learning Based Orthopedic Expert Advisory System Using Decision Tree and Back…
185
186
International Journal of Computational Intelligence Research & Applications
7. FUTURE WORK & CONCLUSION By the thorough interaction with the users and beneficiaries the functionality of the System can be extended further to many more areas in and around the world.
[3] “SAMS Teach Your Self UML in 24 Hours”, Joseph Schmuller, Tech media. [4] java technology. [5] Expert System Testing. [6] www.onlinediagnosis.com.
REFERENCES & BIBLIOGRAPHY [1] “Artificial Intelligence Developments and Applications” by J.S.Gem and R.Stranton, North Holland Pub., 1987. [2] “Programming Expert Systems in Pascal” by Brain Sawyer and Dennis Foster, Addision Wesley pub..
[7] www.medindia.com. [8] www.easydiagnosis.com. [9] Artificial Intelligence and Expert Systems by Patterson. [10] Wikipedia, google search.
