BMI Course Information
BMI 221 Knowledge Representation for Biomedical Informatics (3)
Instructor Information |
|
Fall 2009 |
Course not offered |
Spring 2009 |
Course not offered |
Catalog Description
Introduction to topics in knowledge representation and modeling, including frame-based systems, logic-based systems, rule-based systems, inference, and reasoning. As medicine is a knowledge-rich domain, systems designed to support performance of practitioners require some representation of this domain knowledge. This course presents the key formalisms that have been used encode knowledge into historical and contemporary biomedical informatics systems.
Prerequisites
BMI 201
Textbook and Other Materials
Artificial Intelligence: A modern Approach by Stuart Russell and Peter Norvig, Prentice Hall, ISBN: 0-13-103805-2
Optional: “Model Driven Architecture and Ontology Development” by Dragon Gasevic, Dragan DJuric, and Vladan Devedzic, Springer ISBN: 3-540-32180-2
Course Learning Outcomes
Students who complete this course will be able to:
1. Identify and describe the three major representational languages.
2. Identify and describe ways of representing uncertainty.
3. Critically appraise existing applications and methods in this areas.
4. Place existing applications in the context of the history of the field.
5. Describe current medical ontologies and medical vocabularies.
Major Topics and Time Covered
Introduction to Knowledge representation (2 weeks)
- Concept map
- Conceptual graphs
- Symbolic representation
Logic-based representational language
- First order logic (1 week)
- Predicate logic (1 week)
- Propositional logic (1 week)
- Clinical inference rules
Frame-based representation language (protégé) (2 weeks)
- Biomedical Ontologies
Rule-based representation languages (prolog) (2 weeks)
- Historical perspective (Mycin)
Representing uncertainty (2 weeks)
- Uncertainty in biomedical problem solving
- Fuzzy logic
- Fuzzy sets
- Certainty factors
- Bayesean logic
- Dempster-Shafer theory
Graph theory representations (1 week)
- Belief networks
- Bayesian networks
- Semantic networks
Medical Knowledge representation systems (2 weeks)
- SNOMED, Gene Ontology, LOINC, UMLS