Computer-Based Medical Consultations: MYCIN

Computer-Based Medical Consultations: MYCIN focuses on MYCIN, a novel computer-based expert system designed to assist physicians with clinical decisions concerning the selection of appropriate therapy for patients with infections. It discusses medical computing, artificial intelligence, and the clinical problem areas for which the MYCIN program is designed, and it describes in detail how the MYCIN program helps physicians in making decisions.
Comprised of seven chapters, this volume begins with an overview of MYCIN and the criteria used in its design. It then discusses data structures and control structures in the context of prior work regarding rule-based problem-solving, inferential model building and inexact reasoning in medicine. The book also explores MYCIN'S ability to answer questions with respect to its knowledge base and the details of a specific consultation, evaluation and future extensions of the MYCIN system, the limitations and accomplishments of MYCIN, and its contributions in artificial intelligence and computer-based medical decision making.
This book is a valuable source of information for computer scientists and members of the medical community.

Contenu

Preface
Foreword
Chapter 1. Introduction
1.1 Computer Applications in Medicine
1.1.1 Problems and Promises
1.1.2 Medical Computing Application Areas
1.2 Artificial Intelligence
1.2.1 Areas of Application
1.2.2 AI Methodologies and Techniques
1.3 Computer-Assisted Medical Decision Making
1.3.1 Major Problem Area
1.3.2 Data Retrieval as a Decision Aid
1.3.3 Decisions Based on Numerical Computations
1.3.4 Probabilistic Approaches to Decision Making
1.3.5 Artificial Intelligence and Medical Decisions
1.3.6 Philosophical Observations
1.4 Antimicrobial Selection
1.4.1 Nature of the Decision Problem
1.4.2 Evidence that Assistance is Needed
1.5 MYCIN System
1.5.1 System's Organization
1.5.2 Sample Consultation Session
Chapter 2. Design Considerations for Mycin
2.1 Introduction
2.2 Design Considerations for Consultation Programs
2.2.1 Program Should be Useful
2.2.2 Program Should be Educational when Appropriate
2.2.3 Program Should be Able to Explain Its Advice
2.2.4 Program Should be Able to Understand Questions
2.2.5 Program Should be Able to Acquire New Knowledge
2.2.6 Program's Knowledge-Base Should be Easily Modified
2.3 MYCIN and Acceptability Criteria
2.3.1 Modularity to Insure Straightforward Modification
2.3.2 Ability to Acquire New Knowledge from Experts
2.3.3 Ability to Understand Questions
2.3.4 Ability to Explain Decisions
2.3.5 Educational Capabilities
2.3.6 General Usefulness
Chapter 3. Consultation System
3.1 Introduction
3.2 System Knowledge
3.2.1 Decision Rules
3.2.2 Categorization of Rules by Context
3.2.3 Clinical Parameters
3.2.4 Certainty Factors
3.2.5 Functions for Evaluation of Premise Conditions
3.2.6 Static Knowledge Structures
3.2.7 Translation of Rules into English
3.3 Use of Rule s to Give Advice
3.3.1 Previous Goal-Oriented Problem Solving Systems
3.3.2 MYCIN'S Control Structure
3.3.3 Creation of Dynamic Data Base
3.3.4 Self-Referencing Rules
3.3.5 Preventing Reasoning Loops
3.4 Propagation of Context Tree
3.4.1 Data Structures Used for Sprouting Branches
3.4.2 Explicit Mechanisms for Branching
3.4.3 Implicit Mechanisms for Branching
3.5 Selection of Therapy
3.5.1 Creation of Potential Therapy List
3.5.2 Selecting Preferred Drug from List
3.6 Mechanism s for Storage of Patient Data
3.6.1 Changing Answers to Questions
3.6.2 Remembering Patients for Future Reference
3.7 Future Extension s
3.7.1 Dynamic Ordering of Rules
3.7.2 Dynamic Ordering of Conditions within Rules
3.7.3 Pre-Screening of Rules
3.7.4 Placing all Knowledge in Rules
3.7.5 Need for Context Graph
3.8 Advantage s of MYCIN Approach
3.8.1 Modularity of Knowledge
3.8.2 Dynamic Reasoning Chain
3.8.3 Domain-Independent Control Structure
3.8.4 Reasoning with Judgmental Knowledge
Chapter 4. Model of Inexact Reasoning in Medicine
4.1 Introduction
4.2 Problem Formulation
4.3 MYCIN'S Rule-Based Approach
4.4 Theoretical Background
4.5 Proposed Model of Evidential Strength
4.6 Model as Approximation Technique
4.7 MYCIN'S Use of Model
Chapter 5. Explanation System
5.1 Introduction
5.2 Using Question-Answering System
5.2.1 Rule-Retrieval Questions
5.2.2 Questions Regarding Dynamic Base Data
5.2.3 Additional Options
5.3 Future Extensions
Chapter 6. Future Directions for Mycin
6.1 Introduction
6.2 Plans for Immediate Future
6.3 Knowledge Acquisition
6.3.1 Current Status of Rule-Acquisition
6.3.2 Future Extensions
6.4 Evaluation of MYCIN
6.4.1 Reliability of MYCIN'S Advice
6.4.2 MYCIN'S Acceptability to Physicians
6.4.3 MYCIN'S Impact on Prescribing Habits
6.4.4 MYCIN'S Impact on Patient Care
6.4.5 Speed, Efficiency, and Storage Requirements
6.4.6 Cost of MYCIN'S Consultations
6.5 MYCIN and Shared Data Bases
6.6 Prospective Monitoring of Prescribing Habits
6.7 Educational Applications
6.8 Other Applications of MYCIN Formalism
Chapter 7. Conclusion
7.1 Summary
7.1.1 The Clinical Problem
7.1.2 The Solution
7.2 Limitation s of MYCIN' S Approach
7.3 Contributions to Computer-Based Medical Decision Making
7.4 Contribution to AI
References
Index