Font Size: a A A

Using decision models to automate and individualize interactive decision support for patients

Posted on:2004-12-22Degree:Ph.DType:Dissertation
University:Stanford UniversityCandidate:Scott, George ChristopherFull Text:PDF
GTID:1468390011974686Subject:Computer Science
Abstract/Summary:
I created a novel approach for allowing patients to learn more about their values, and the impact their values have on therapeutic decisions, through automated decision analytic conversations. I proposed using a generic decision model to guide an automated, interactive decision analytic conversation following a dynamically generated path. The conversation results in the customization of the model to an individual's quantities. This method yields recommendations consistent with the individual's quantities, is efficient with respect to the number of assessments required, and is more robust to measurement error than a pre-determined ordering of quantity assessments. The interaction and the report automatically generated on information learned during the conversation are acceptable to users and they would use such a system for their own medical decisions.; I developed a framework called SecondOpinion for providing interactive decision support guided by a decision model. In order to manage the interaction, I created a model for an automated decision analytic conversation. To guide the discussion, I developed an algorithm for intelligently guiding that conversation by individualizing a generic decision model. To communicate the results of the conversation, I created a computer-generated report to provide insight about the decision to both the patient and their healthcare provider. I describe the software architecture I designed to implement the SecondOpinion framework and demonstrate examples for different decision contexts.; I assessed the framework's ability to obtain accurate customizations of the decision models efficiently and with robustness to the level of noise in assessments. The SecondOpinion approach was more accurate, efficient, and robust than the standard approach of a pre-determined assessment of all quantities for all four decision contexts examined. Second, I evaluated the user acceptance of and satisfaction with a SecondOpinion framework decision aid compared to a standard full assessment decision aid. SecondOpinion was as acceptable as the full assessment aid, and study participants were as satisfied with the new framework as with the full assessment approach.; I concluded that the use of approaches such as the SecondOpinion framework have the potential of improving the quality of healthcare while making the delivery of individualized healthcare more efficient.
Keywords/Search Tags:Decision, Secondopinion framework, Approach
Related items