A therapy planning architecture that combines decision theory and artificial intelligence techniques.

Area of Science:

• Artificial Intelligence in Medicine
• Computational Planning
• Decision Theory

Background:

• Existing computer-based reasoning techniques are insufficient for complex medical planning problems.
• Experience with the ONCOCIN cancer therapy consultation system highlighted these limitations.

Purpose of the Study:

• To devise a novel computer program, ONYX, for automated assistance in medical planning.
• To combine decision-theoretic and artificial intelligence approaches for enhanced planning capabilities.

Main Methods:

• Developed a new planning architecture implemented in the ONYX program.
• Generated plausible plans using therapy planning rules.
• Simulated plan consequences using human body knowledge.
• Ranked plans using decision theory based on simulation outcomes.

Main Results:

• The ONYX architecture explicitly manages uncertainty in planning tasks.
• It provides a mechanism for disseminating decision-theoretic therapy advice.
• It expands the applicability of expert system techniques to new problem domains.

Conclusions:

• The ONYX system offers a robust approach to complex medical planning.
• This integrated methodology enhances automated decision support in healthcare.
• Further application of this architecture can improve expert system utility in diverse fields.