Medical Diagnosis

1. A Natural Language based Knowledge Representation Method for Medical Diagnosis Dr Peter Scholten Dr Jie Ji, Huilin Chen and Yixuan Song SAI Computing Conference 2016 July 13-16, 2016 | London UK

2. What is a medical diagnose? An identification of a disease based on objective ( measurable ) and /or subjective ( not measurable ) symptoms . A subjective symptom is a description of the feelings by a patient , like: I have fever , I have pain , I’m tired .

3. Misdiagnose > 5%

4. What is the problem? Reliable data Classification for diseases , but no classification for symptoms System based on natural language Different users: - medical doctors and healthconsumers - different cognitive levels Missing direct interaction between doctor and patient

5. Non treatment Treatment Definition of diseases 1

6. Diseases Symptoms hasSymptom isSymptomOf An inverse property: hasSymptom is inverse of isSymptomOf class class

7. diseases Symptom 1 Symptom 3 Symptom 2 RDF: ( s,p,o ) triple

8. disease Symptom 1 Symptom 3 Symptom 2 isSymptomOf hasSymptom hasSymptom

9. disease Select symptom Candidate symptom Candidate symptom

10. 1625 9.723 56.912 / 10.230 / 400

11. User input conjugate Concept disease Select symptom Candidate symptom Candidate symptom Y N

12. The l inguistic ontology

13. Human factors effect the accuracy 1 . The errors due to accuracy of the content. 2. The errors due to interoperability of the application. 3. The errors due to shortcomings of the user . illness ability to describe illness

14. = n umber of symptoms Description of symptoms

15. Results: The system does work like a human expert, which is able to read and learning , and understand natural language inputs .

16. Demo

17. Heart attack and personal conditions

18. : circulatory

19. Heart attack Pressure pain chest and neck Personal conditions Age > 45 Hereditary YES Active NO Overweight YES Smoking YES

20. Contact [email protected]