| Abstract
|
This dissertation addresses automated assi … This dissertation addresses automated assistance fordecision analysis in medicine. In particular, I have investigated graphgrammars as a representation for encoding how decision-theoretic models can beconstructed from an unordered list of concerns. The modeling system that Ihave used requires a standard vocabulary to generate decision models; themodels generated are qualitative, and require subsequent assessment ofprobabilities and utility values. This research has focused on the modeling ofthe qualitative structure of problems given a standard vocabulary and giventhat subsequent assessment of probabilities and utilities is possible. Theusefulness of the graph-grammar representation depends on the graph-grammarformalism's ability to describe a broad spectrum of qualitative decisionmodels, on its ability to maintain a high quality in the models it generates,and on its clarity in describing topological constraints to researchers whodesign and maintain the actual grammar. I have found that graph grammars canbe used to generate automatically decision models that are comparable to thoseproduced by decision analysts. ble to thoseproduced by decision analysts.
|
