DRS Presentation
date post
19-Feb-2017Category
Documents
view
228download
2
Embed Size (px)
Transcript of DRS Presentation
Diabetes Recommender System
Diabetes Recommender System)DRS)(Using Semantic Web and Data Mining techniques)7/5/2015DRS 1
Nesma Mahmoud Saad EddinNada Hassan NaemBasma Gamal El-serafyHanan Sabery Hamed Menna Talla Mamdouh Saoud
Subervised by Dr. Heba ElbehTeam Members7/5/2015DRS 2
What is DRS?Why is DRS?System Architecture & ComponentsDiabetes Diagnosis (using DM)Drug & Diet Recommendation (using SW)How DRS WorksDRS Tools & Techniques
Agenda7/5/2015DRS 3
DRS (Diabetes recommender system) is a web-based recommendation system for diabetes. It provides diagnosis for diabetes, recommends diabetics drugs, recommends and advice diabetes.1.What is DRS?
7/5/2015DRS 4
DRS is based on Data Mining and Semantic Web techniques provides more accuracy in results.DRS provides more than one service (diabetes diagnosis, drug recommendation, diet recommendation) DRS maintain personalization for users(pateints).Diabetes is one of the most chronic diseases in recent years.The number of drugs increased the number of patients increased.
2. Why DRS?
7/5/2015DRS 5
3. DRS Architecture and Components. 7/5/2015DRS 6
4. Diabetes Diagnosis(Using Data Mining)7/5/2015DRS 7
Data Mining Data Mining Steps
7/5/2015DRS 8
Data Collection & Preprocessing - Diabetes Database from UCI Machine Learning . - It Contains 768 record samples records with 8 attributes (e.g., age, number of pregnant, Plasma Glucose test, etc.) - Pre-processing replacing missing values, convert numeric values to nominal, discretize attributes.Model Building technique - We use decision tree for data mining classification. - we use J48 Classifier for building the model(Implementation of Decision tree).Evaluation & Deployment - Accuracy of the classifier is about 75.78%.
Data Mining Steps7/5/2015DRS 9
J48 Classifier is training and testing the preprocessed diabetics data set. Producing a model in the form of tree and we use this model in our DRS system for Diagnosing new users (diabetic or not)The resulting classification tree model:Classifier Model Building
7/5/2015DRS 10
5. Drugs & Diet Recommendation(Using Semantic Web)7/5/2015DRS 11
Evolving extension of WWW in which web content : - Can be expressed not only in natural language. - But also can be understood, interpreted and used by software agents. - Permitting software agents to find, share and integrate information easily.The idea of having data on the Web - Defined and linked in such a way. - Can be used by machines not only for display purposes, but for automation, integration and reuse of data across various applications
Semantic Web (SW)
7/5/2015DRS 12
Semantic Web Cake Layer.SW Knowledge Representation & Layers
Ontology7/5/2015DRS 13
Why Ontologies are important - Ontological analysis clarifies the structure of knowledge. - Ontologies enable knowledge sharing. Ontology Definition: Formal , explicit specification of a shared conceptualization
Ontology
Machine readable
Concepts, properties,functions, axiomsare explicitly defined
Consensualknowledge
Abstract model of some phenomenain the world7/5/2015DRS 14
Ontology Main Components
Ontology (cont.)PersonCountry
Class (concept)
Animal
Individual (instance)
BelgiumParaguayChinaLatvia
ElvisHaiHolgerKylieS.Claus
Rudolph
Flipperarrow = relationshiplabel = Property
lives_inlives_inlives_in
has_pethas_pethas_pet7/5/2015DRS 15
Step by Step Building Ontology7/5/2015DRS 16
Anti-diabetic drugs ontology (an ontology that maintain ant-diabetic drugs).Patient tests ontology ( an ontology that maintain diabetes tests).Diabetes food ontology (an ontology that maintain all foods for diabetics diet). Personal Patient Information ( ontology for patients personal and medical information)DRS Ontologies7/5/2015DRS 17
Define Scope: anti-diabetics drugs ontology for drug recommendation for diabetics Patients in DRS system.Data Gathering & Consider reuse: reusing drugs ontology, collecting data from internet and asking specialists.Enumerate Terms: List all nouns and verbs used in the domain like (drug, drugs classes, drugs names, has consideration, has dose time, etc.)Define Classes: nouns become classes in the ontology.
Anti Diabetic Drugs Ontology7/5/2015DRS 18
Define classes hierarchy: define classes in taxonomic (subclass) hierarchy.
Anti Diabetic Drugs Ontology(cont.)
7/5/2015DRS 19
Define Properties: define relationships . Data property: link individual to individual.Object property: link individual to literal.
Anti Diabetic Drugs Ontology(cont.)
7/5/2015DRS 20
Define Scope: represents diabetics patient tests and used for drug recommendation with the drug ontology in DRS system.Data Gathering & Consider reuse: reusing patients tests ontology, collecting data from internet and asking specialists.Enumerate Terms: List all nouns and verbs used in the domain like (test, test name, test classes, has normal has level, etc.).Define Classes: nouns become classes in the ontology.
Patient Tests Ontology7/5/2015DRS 21
Define classes hierarchy: define classes in taxonomic (subclass) hierarchy.
Patient Tests Ontology(cont.)
7/5/2015DRS 22
Define Scope: represents foods that allowed for diabetic patients and knowledge about foods and used for diet recommendation in DRS system.Data Gathering & Consider reuse: reusing foods ontology, collecting data from internet and asking specialists.Enumerate Terms: List all nouns and verbs used in the domain like (test, test name, test classes, has normal has level, etc.).Define Classes: nouns become classes in the ontology.
Diabetes Foods Ontology7/5/2015DRS 23
Define classes hierarchy: define classes in taxonomic (subclass) Diabetes Foods Ontology (cont.)
7/5/2015DRS 24
Define Properties: define relationships . Data property: link individual to individual.Object property: link individual to literal.
Diabetes Foods Ontology (cont.)
7/5/2015DRS 25
Define Scope: represents all patient information from personal information to drug and diet information that use to manage in DRS system.Data Gathering & Consider reuse: reusing ontology personal ontologies and gathering all data needed for a patients.Enumerate Terms: List all nouns and verbs used in the domain like (patient name, age , allowed calories, etc.)Define Classes: nouns become classes in the ontology.
Patient Information Ontology7/5/2015DRS 26
Define classes hierarchy: define classes in taxonomic (subclass) hierarchy.Patient Information Ontology
7/5/2015DRS 27
SWRL Semantic Web Rule LanguageSWRL is a rule language for semantic webAll rules are expressed in terms of OWL concepts (classes, properties, individuals).Rule Example:
We use in writing the diabetes medication rules.
SWRL for Medication Rules7/5/2015DRS 28
We used the SWRL in DRS to represent the relationship between diabetics important tests and suitable drugs. Diabetics medication rules:
7/5/2015DRS 29SWRL for Medication Rules
6. How DRS Works?
7/5/2015DRS 30
DRS Diabetes Diagnosis
Users
User Interface FormWeka API1.Get user information2. Load Model3. Classify User4. Return Result back to userClassifier Model- Stored in DRSLoading modelSubmitCheck resultPassing userdataModel Result 7/5/2015DRS 31
DRS Drug recommendation
Protg API7/5/2015DRS 32Jess EnginePatient inputRecommended Drug or patientDiabeticsUsers
SubmitBridge
Drugs Ontology
Patients Tests ontologySWRL Medication RulesLoading
DRS Diet Recommendation7/5/2015DRS 33
Diabetics Users
SQWRLQueriesFoods OntologyPatients Information OntologyProtg APIJess EngineRecommended Personal meal BridgeLoadingFavorite Foods& Personal Information
7/5/2015DRS 347. DRS Tools & Techniques
Thanks For Listening 7/5/2015DRS 35
Questions? 7/5/2015DRS 36