Reasoning about Causes with Preferences and Uncertainty in Dynamic Biomedical Domains Reasoning about Causes with Preferences and Uncertainty in Dynamic Biomedical Domains Diagnosis and treatment of diseases or disorders is a complex process, often requiring physicians to consider multiple causes with relative likelihoods and uncertain eects. Any biomedical Question- Answering (QA) system that is required to assist physicians in these tasks needs to be able to encode and reason with such complex knowledge. Consider the sentence \Heart failure is usually due to myocardial damage but may also be the consequence of arrhythmia or mechanical complications such as mitral regurgitation or ventricular septal defect". This sentence indicates that heart failure has more than one cause and that usually, the cause is myocardial damage. Further, the statement also indicates that myocardial damage, arrhythmia, mitral regurgitation and ventricular septal defect \may"cause heart failure. We expect a QA system to be able to recognize myocardial damage as the most likely cause of heart failure, but, at the same time, be able to return other causes as well. Furthermore, we expect the system to be able to predict that myocardial damage, arrhythmia, mitral regurgitation and ventricular septal defect \might"cause heart failure. The system can accomplish the former by encoding some kind of preference between the causes, and the latter by encoding the uncertainty in the eects of the causes. This proposal is about providing a logic programming based approach for representing and reasoning with preferences and uncertainty in dynamic biomedical domains. In particular, we propose to extend Answer Set Programming (ASP) [Lifschitz, 2008] to allow representation of preferences and uncertainty. Since dynamic domains can be represented in ASP, this new language will be able to address all the requirements stated above. The proposed research will provide an opportunity for the collaboration between ASU and Siemens Research Lab, and will mutually benet each other.
|Effective start/end date||9/1/11 → 12/31/11|
- INDUSTRY: Domestic Company: $24,045.00
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.