An extensible six-step methodology to automatically generate fuzzy DSSs for diagnostic applications

Area of Science:

• Computer Science
• Artificial Intelligence
• Medical Informatics

Background:

• Disease diagnosis is often a decision problem complicated by uncertainty.
• Computerized Diagnostic Decision Support Systems (DDSSs) aim to improve clinical data interpretation.
• Fuzzy logic offers a robust approach to handle noisy information in decision-making.

Purpose of the Study:

• To formalize and refine a general methodology for automatically building DDSSs.
• To develop a modular and portable software architecture for implementing the methodology.
• To demonstrate the feasibility of the approach through a proof-of-concept application.

Main Methods:

• A six-stage methodology, with the first three stages using crisp rules and the last three employing fuzzy models.
• A component-based software architecture designed for generality and modularity.
• Implementation in Java using an object-oriented paradigm.

Main Results:

• A modular and portable software architecture was designed and implemented.
• A proof-of-concept DDSS for diagnosing breast masses was successfully instantiated.
• The architecture facilitates the integration of alternative techniques within its stages.

Conclusions:

• The developed methodology and software architecture are feasible for creating automated DDSSs.
• The approach shows promise for enhancing diagnostic accuracy and efficiency.
• The modular design allows for flexibility and future enhancements.