1988 Fiscal Year Final Research Report Summary
On the Development of New Software Algorithms for the New Generation Computer-aided Diagnosis
| Project/Area Number |
60440043
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (A)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
内科学一般
|
|---|
| Research Institution | University of Tokyo |
|---|
Principal Investigator |
FURUKAWA Toshiyuki Faculty of Medicine, University of Tokyo, Professor, 医学部, 教授 (20101082)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
KAIHARA Shigekoto Faculty of Medicine, University of Tokyo, Professor, 医学部, 教授 (30010234)
SUGIMOTO Tuneaki Faculty of Medicine, University of Tokyo, Professor, 医学部, 教授 (60019883)
|
|---|
| Project Period (FY) |
1985 – 1988
|
| Keywords | New generation computer / New generation computer-aided diagnosis / hierarchical inference logic / Artificial intelligence / Knowledge formation / Ambiguity of biological data / Disease process model / Inverse solution / 不完全データ |
|---|
| Research Abstract |
In order to develope the new software algorithms for the new generation computeraided diagnosis, a series of study for 4 years was completed.
(1) Hierarchical tree structure logics for the clinical decision making were studied using the real cases of infusion therapy. It showed the decision capability fairly resembled to that of physicians.
(2) Multivariate statistical techniques using the incomplete data cources were studied. The study revealed the efficacy of the maximum entropy method to substitute the lack of necessary data.
(3) The clinical courses of disease were simulated by a network with nodes correspond to the states of the disease. These models showed good agreement with the data of clinical epidemiological studies. There should be a certain minimum spacial and time distance of clinical examinations in order to describe the clinical course of the diseases. The proposal is the application of multivariate statistical technique to determing the spacial distance, and the Markov process model to determine the time distance of the examination.
(4) A parallel data processing using a number of microcomputers was introduced in the case of an expert system to the infusion therapy, and the communication and data exchange between the different computers with different operation systems and/or different computer languages could be effectively coordinated. In the study, the simulation model of body fluid regulation physiology was implemented in a computer, and the decision logic of an artificial intelligence was equipped into another computer. In addition, the process of learning of physician was analyzed by a multivariate statistical model, and the community between them was found. It is fundamental condition to give an appropriate data base (teaching) at very early stage of learning is the most important factor in not only the maturation of physician's knowledge but that of computer-aided diagnostic logics.
|
