Computer Assisted Medical Diagnosis Problems and Methods to Minimize Their Effects.
Abstract
The background and methods used in Computer Assisted Medical Diagnosis (CAMD) are discussed and the problems and errors encountered during the development of CAMD systems are examined. Five specific methods and algorithms are discussed in terms of the problems previously reported in the literature, as well as, possible solutions and ways to minimize these problems. The use of multiple experts, adequate sampling sizes, multicenter samples with adequate and representative disease cases in the development of CAMD systems is recommended. Limitations of these systems can be determined from thorough system validation and testing of disease cases and sign/ symptom data that were not used in the initial system development. The utilization of sign and symptom disease complexes (indice, scales), and the use of multiple disease diagnostic methods to identify possible disease diagnosis is emphasized. Furthermore, CAMD systems should give standard definitions and interpretations for any signs, symptoms, lab results, diseases, or treatments suggested, and at a time the level of diagnostic expertise of the user should be considered. The development of the CAMD systems using a data base management approach that allow the integration of multiple methods is discussed as a valuable approach for implementing systems. Computer Assisted Medical Diagnosis, diagnostic algorithms Expert systems, Rule-based, Neural Network.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1991
- Accession Number
- ADA258367
Entities
People
- David H. Ryman
Organizations
- Naval Health Research Center