The Heart Failure Program project is a resource-related research program to develop new reasoning methods for the application of Artifical Intelligence techniques to medicine for the effort of the SUMEX-AIM community. The context and driving force for this research is the management of heart failure in the intensive care setting. We will: 1) Develop a representational methodology capable of supporting the clinically relevant distinctions of patient state. This representation will utilize the clinically significant qualitative parameter values and will include causal relationships, time dependencies and relations about change. 2) Build a qualitative physiological model of the cardiovascular system using this methodology to act as a store for evolving knowledge of patient state. 3) Explore and develop strategies for determining the appropriate parameter values in the model from input data, reasoning support methods and heuristics for carrying out the diagnostic reasoning with the model, and reasoning support methods and heuristics for carrying out the diagnostic reasoning with the model, and reasoning support methods and heuristics for finding possible therapies and determining their potential consequences. 4) Build around this core a program to assist the physician in exploring his or her understanding of the implications in an individual case. The physician and program will reason together about the case with the physician providing ideas and the program assuring consistent consideration of the implications. 5) Generalize the techniques for use in other medical domains.

National Institute of Health (NIH)
National Heart, Lung, and Blood Institute (NHLBI)
Research Project (R01)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG7-SSS-9 (11))
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Massachusetts Institute of Technology
Organized Research Units
United States
Zip Code
Fraser, Hamish S F; Long, William J; Naimi, Shapur (2003) Evaluation of a cardiac diagnostic program in a typical clinical setting. J Am Med Inform Assoc 10:373-81
Long, W J (2001) Medical informatics: reasoning methods. Artif Intell Med 23:71-87
Fraser, H S; Naimi, S; Long, W J (2000) New approaches to measuring the performance of programs that generate differential diagnoses using ROC curves and other metrics. Proc AMIA Symp :255-9
Fraser, H S; Long, W J; Naimi, S (1998) Differential diagnoses of the heart disease program have better sensitivity than resident physicians. Proc AMIA Symp :622-6
Ohno-Machado, L; Fraser, H S; Ohrn, A (1998) Improving machine learning performance by removing redundant cases in medical data sets. Proc AMIA Symp :523-7
Long, W J; Fraser, H; Naimi, S (1997) Reasoning requirements for diagnosis of heart disease. Artif Intell Med 10:5-24
Long, W (1996) Temporal reasoning for diagnosis in a causal probabilistic knowledge base. Artif Intell Med 8:193-215
Long, W J; Fraser, H; Naimi, S (1996) Web interface for the Heart Disease Program. Proc AMIA Annu Fall Symp :762-6
Long, W J; Naimi, S; Criscitiello, M G (1994) Evaluation of a new method for cardiovascular reasoning. J Am Med Inform Assoc 1:127-41
Long, W J; Griffith, J L; Selker, H P et al. (1993) A comparison of logistic regression to decision-tree induction in a medical domain. Comput Biomed Res 26:74-97

Showing the most recent 10 out of 19 publications