I build an expert system design that applies a combination of Certainty Factor and K-Nearest Neighbor methods in solving problems of diagnosing dental diseases (especially for teledentistry). The expert system is website-based (using Jinja 3.0 and Bootstrap AdminLTE.io) and uses MySQL. The application provides an accuracy of 80% in the Certainty factor-only method using result validation and the combined accuracy of the Certainty Factor - K Nearest Neighbor method of 90% using Cross Validation (to see the calculation of 90% accuracy: https://github.com/fanimasturina/CF-KNN-Accuracy-Using-Jupyter-Notebook). The dataset is used on combination methods to serves the neighbor classification, the detail of dataset is consist of 100 patients medical records (name, certainty value of symptomps, and diagnosis).

How to run the app on local: https://github.com/fanimasturina/Expert-System-Certainty-Factor-K-Nearest-Neighbor-Dental-Desease/blob/main/HowToRunOnLocal.md

The preview of an app:

1. Homepage

2. Disease Info Page

3. Symptoms Info Page

4. Patients Medical Record Page

5. Knowledge Base Page

6. Choosing Symptoms Page

7. Patient Complaints Page

8. Diagnosis Results using Certainty Factor method only

9. Diagnosis Results using combination of Certainty Factor & K Nearest Neighbor methods

To read more about the projects:

• Working paper: https://github.com/fanimasturina/Expert-System-Certainty-Factor-K-Nearest-Neighbor-Dental-Desease/blob/main/Working-Paper_2001573676.doc
• Jurnal Ilmiah: https://github.com/fanimasturina/Expert-System-Certainty-Factor-K-Nearest-Neighbor-Dental-Desease/blob/main/Jurnal-Ilmiah_2001573676.doc