Diagnostics and a qualitative model.

Area of Science:

• Artificial Intelligence
• Computer Science
• Medical Diagnostics

Background:

• First-generation expert systems used shallow, heuristic knowledge for diagnostics, presenting significant limitations.
• Deep knowledge, particularly through qualitative modeling, offers a more robust approach to diagnostic reasoning.
• Model-based diagnostics leverages deep knowledge for enhanced problem-solving capabilities.

Purpose of the Study:

• To present a formal model-based diagnostic approach.
• To introduce a novel formalization including diagnostic space, minimal diagnoses, and measurement.
• To establish the DISY diagnostic architecture based on qualitative system models.

Main Methods:

• Formalization of model-based diagnostics.
• Development of a diagnostic space concept.
• Design of the DISY diagnostic architecture utilizing qualitative system models.
• Simulation of system behavior under normal and abnormal functioning.

Main Results:

• A formal framework for model-based diagnostics was established.
• The concept of diagnostic space and minimal diagnoses were characterized.
• The DISY architecture was proposed, capable of integrating previous diagnostic results with new observations.
• Qualitative system models were shown to be adaptable for diagnostic purposes.

Conclusions:

• The presented formalization provides a theoretical foundation for model-based diagnostics.
• The DISY architecture offers a computationally efficient approach to diagnostics by incorporating sequential observations.
• Qualitative system models are versatile and do not require domain-specific adaptation for diagnostic applications.