An Epistemological Framework for Medical Knowledge-Based Systems

Marco Ramoni, Lorenzo Magnani, Giovanni Barosi

Research output: Contribution to journalArticle

66 Citations (Scopus)

Abstract

There is a general agreement about the need of identifying a highlevel conceptual design in knowledge-based systems that is behind the implementation. A number of perspectives for understanding knowledge-based systems at this high level has been proposed. Major examples of these perspectives are based on the concepts of heuristic classification, deep-shallow systems, problem-solving method, and generic task. Each of these ideas focuses on a particular feature of reasoning: the inference structure, the models of domains knowledge, the sequence of actions need to solve a problem, and the task features. This paper presents a new abstraction paradigm aiming at unifying these different perspectives. The proposed model should be able to account for all of the conceptual features of knowledge-based systems, thus making clear which features are intrinsic to the problem and which are artifacts of the implementation. The proposal is therefore based on a two-level analysis of knowledge-based systems in medical reasoning: an epistemological and a computational level. At the first level, ontology and inference model of a knowledge-based system have to be defined. Ontology represents the conceptual model of domain knowledge, while the inference model is the conceptual representation of the inference structure needed to solve a problem or to execute a task by managing that ontology. At the computational level, methods and formalisms should be adopted after the epistemological analysis has been carried out and taking into account the constraints that derive from the conceptual structure of the domain knowledge, patterns of inference, and tasks to be executed. The study was constrained to medicine and identified three generic tasks: diagnosis, therapy planning, and monitoring. The main result of this analysis is that these generic tasks manage different ontologies, but can be executed exploiting a unique inference model. Such a model involves three different inference types (abduction, deduction, and induction) and is described in some details. Finally, computational issues are discussed to argue that the present model provides a conceptual view on existing systems, and some design insights for future ones.

Original languageEnglish
Pages (from-to)1361-1375
Number of pages15
JournalIEEE Transactions on Systems, Man and Cybernetics
Volume22
Issue number6
DOIs
Publication statusPublished - 1992

Fingerprint

Knowledge based systems
Ontology
Conceptual design
Medicine
Planning
Monitoring

ASJC Scopus subject areas

  • Engineering(all)

Cite this

An Epistemological Framework for Medical Knowledge-Based Systems. / Ramoni, Marco; Magnani, Lorenzo; Barosi, Giovanni.

In: IEEE Transactions on Systems, Man and Cybernetics, Vol. 22, No. 6, 1992, p. 1361-1375.

Research output: Contribution to journalArticle

Ramoni, Marco ; Magnani, Lorenzo ; Barosi, Giovanni. / An Epistemological Framework for Medical Knowledge-Based Systems. In: IEEE Transactions on Systems, Man and Cybernetics. 1992 ; Vol. 22, No. 6. pp. 1361-1375.
@article{7c2d7cfa8a1448a3a5f5717d9983816a,
title = "An Epistemological Framework for Medical Knowledge-Based Systems",
abstract = "There is a general agreement about the need of identifying a highlevel conceptual design in knowledge-based systems that is behind the implementation. A number of perspectives for understanding knowledge-based systems at this high level has been proposed. Major examples of these perspectives are based on the concepts of heuristic classification, deep-shallow systems, problem-solving method, and generic task. Each of these ideas focuses on a particular feature of reasoning: the inference structure, the models of domains knowledge, the sequence of actions need to solve a problem, and the task features. This paper presents a new abstraction paradigm aiming at unifying these different perspectives. The proposed model should be able to account for all of the conceptual features of knowledge-based systems, thus making clear which features are intrinsic to the problem and which are artifacts of the implementation. The proposal is therefore based on a two-level analysis of knowledge-based systems in medical reasoning: an epistemological and a computational level. At the first level, ontology and inference model of a knowledge-based system have to be defined. Ontology represents the conceptual model of domain knowledge, while the inference model is the conceptual representation of the inference structure needed to solve a problem or to execute a task by managing that ontology. At the computational level, methods and formalisms should be adopted after the epistemological analysis has been carried out and taking into account the constraints that derive from the conceptual structure of the domain knowledge, patterns of inference, and tasks to be executed. The study was constrained to medicine and identified three generic tasks: diagnosis, therapy planning, and monitoring. The main result of this analysis is that these generic tasks manage different ontologies, but can be executed exploiting a unique inference model. Such a model involves three different inference types (abduction, deduction, and induction) and is described in some details. Finally, computational issues are discussed to argue that the present model provides a conceptual view on existing systems, and some design insights for future ones.",
author = "Marco Ramoni and Lorenzo Magnani and Giovanni Barosi",
year = "1992",
doi = "10.1109/21.199462",
language = "English",
volume = "22",
pages = "1361--1375",
journal = "IEEE Transactions on Systems, Man and Cybernetics",
issn = "0018-9472",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "6",

}

TY - JOUR

T1 - An Epistemological Framework for Medical Knowledge-Based Systems

AU - Ramoni, Marco

AU - Magnani, Lorenzo

AU - Barosi, Giovanni

PY - 1992

Y1 - 1992

N2 - There is a general agreement about the need of identifying a highlevel conceptual design in knowledge-based systems that is behind the implementation. A number of perspectives for understanding knowledge-based systems at this high level has been proposed. Major examples of these perspectives are based on the concepts of heuristic classification, deep-shallow systems, problem-solving method, and generic task. Each of these ideas focuses on a particular feature of reasoning: the inference structure, the models of domains knowledge, the sequence of actions need to solve a problem, and the task features. This paper presents a new abstraction paradigm aiming at unifying these different perspectives. The proposed model should be able to account for all of the conceptual features of knowledge-based systems, thus making clear which features are intrinsic to the problem and which are artifacts of the implementation. The proposal is therefore based on a two-level analysis of knowledge-based systems in medical reasoning: an epistemological and a computational level. At the first level, ontology and inference model of a knowledge-based system have to be defined. Ontology represents the conceptual model of domain knowledge, while the inference model is the conceptual representation of the inference structure needed to solve a problem or to execute a task by managing that ontology. At the computational level, methods and formalisms should be adopted after the epistemological analysis has been carried out and taking into account the constraints that derive from the conceptual structure of the domain knowledge, patterns of inference, and tasks to be executed. The study was constrained to medicine and identified three generic tasks: diagnosis, therapy planning, and monitoring. The main result of this analysis is that these generic tasks manage different ontologies, but can be executed exploiting a unique inference model. Such a model involves three different inference types (abduction, deduction, and induction) and is described in some details. Finally, computational issues are discussed to argue that the present model provides a conceptual view on existing systems, and some design insights for future ones.

AB - There is a general agreement about the need of identifying a highlevel conceptual design in knowledge-based systems that is behind the implementation. A number of perspectives for understanding knowledge-based systems at this high level has been proposed. Major examples of these perspectives are based on the concepts of heuristic classification, deep-shallow systems, problem-solving method, and generic task. Each of these ideas focuses on a particular feature of reasoning: the inference structure, the models of domains knowledge, the sequence of actions need to solve a problem, and the task features. This paper presents a new abstraction paradigm aiming at unifying these different perspectives. The proposed model should be able to account for all of the conceptual features of knowledge-based systems, thus making clear which features are intrinsic to the problem and which are artifacts of the implementation. The proposal is therefore based on a two-level analysis of knowledge-based systems in medical reasoning: an epistemological and a computational level. At the first level, ontology and inference model of a knowledge-based system have to be defined. Ontology represents the conceptual model of domain knowledge, while the inference model is the conceptual representation of the inference structure needed to solve a problem or to execute a task by managing that ontology. At the computational level, methods and formalisms should be adopted after the epistemological analysis has been carried out and taking into account the constraints that derive from the conceptual structure of the domain knowledge, patterns of inference, and tasks to be executed. The study was constrained to medicine and identified three generic tasks: diagnosis, therapy planning, and monitoring. The main result of this analysis is that these generic tasks manage different ontologies, but can be executed exploiting a unique inference model. Such a model involves three different inference types (abduction, deduction, and induction) and is described in some details. Finally, computational issues are discussed to argue that the present model provides a conceptual view on existing systems, and some design insights for future ones.

UR - http://www.scopus.com/inward/record.url?scp=0026943429&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0026943429&partnerID=8YFLogxK

U2 - 10.1109/21.199462

DO - 10.1109/21.199462

M3 - Article

VL - 22

SP - 1361

EP - 1375

JO - IEEE Transactions on Systems, Man and Cybernetics

JF - IEEE Transactions on Systems, Man and Cybernetics

SN - 0018-9472

IS - 6

ER -