An expert system design to diagnose cancer by using a new method reduced rule base

Area of Science:

• Medical Informatics
• Computational Biology
• Artificial Intelligence in Medicine

Background:

• A Medical Expert System (MES) was developed to diagnose cancer risk using a reduced rule base.
• The system utilizes 13 distinct symptoms to assess individual cancer risk.
• Traditional methods involve evaluating numerous possibilities (2^13 = 8192), which is time-consuming.

Purpose of the Study:

• To develop an efficient MES for rapid cancer risk diagnosis.
• To enhance diagnostic accuracy by employing a reduced rule base.
• To enable accessible self-testing for cancer risk via dynamic interfaces.

Main Methods:

• A reduced rule base was generated using the two-level simplification of Boolean functions.
• A truth table with 13 inputs and 4 outputs was constructed.
• Boolean Function Minimization was applied to simplify logical functions and reduce computational complexity.

Main Results:

• The developed MES achieved significantly faster diagnoses compared to evaluating all 8192 possibilities.
• Diagnosis speed gains of 100% were observed for breast and renal cancer.
• A 99% gain in diagnosis rate was achieved for cervical and lung cancer, with improved accuracy.

Conclusions:

• Boolean function minimization enhances MES efficiency by reducing the number of rules evaluated.
• The developed system offers faster, more accurate cancer risk assessments.
• User-friendly interfaces on multiple platforms facilitate self-assessment for early cancer detection and potential prevention.