Multi-agent Systems in Medicine

1
Multi-agent Systems in Medicine

• Štepán Urban

2
Content

• Introduction to Multi-agent Systems (MAS)
• What is an Agent?
• Architecture of Agent
• MAS Platforms
• MAS in Medicine
• System for Semantic Search in Medicine
• Management of Organ Transplants
• Simulation of Mitochondrial Metabolism using
  Multi-agents System

3
What is an Agent?

• An agent 1 is an encapsulated computational
  (or physical, even human) system, that is
  situated in some environment, and that is capable
  of flexible, autonomous behaviour in order to
  meet its design objective.

4
Architecture of an Agent
5
MAS Platforms
6
MAS Platforms

• JADE - Java Agent DEvelopment Framework

7
MAS Platforms

• A-Globe

8
Agent System for Semantic Search in Medicine

• An Example of a Query given by a User

9
OnSSA (Ontology-based Semantic Search Agent)

• An ontology-based information retrieval agent
  system in medicine through bio-related literature
  database MEDLINE
• Automates systematic retrieval of literature in
  medicine by utilizing the Semantic Web languages.
• Ontology part is implemented with DAMLOIL
  language and OilEd editor is used partly.
• The abstracts, author, title, journal, and date
  information of document are extracted into XML
  -based input file in order to evaluate the
  relevancy of documents to a query given. In
  addition to the Semantic Web languages, the rest
  of implementation is done in Java.

10
OnSSA System Overview
11
Biomedical Ontology and Search Engines
12
The Process of Information Retrieval
13
Management of Organ Transplants

• Multi-agent system to help to manage the process
  of organ transplants within Spain.
• Agents
• Emergency Coordinator (EC), which is the national
  coordinator of the 0-emergency cases. This is the
  name given to those patients that are waiting for
  an organ and have reached a very critical
  condition, in which their life is at high risk if
  the transplant is not performed in a very short
  period of time.
• Historical Agent (HA), which receives the data of
  all transplants made in Spain. With this
  information it can keep historical files,
  elaborate statistics, apply data mining
  techniques in order to gain useful knowledge,
  etc.

14
Steps in the Search for an Appropriate Recipient.
15
Simulation of Mitochondrial Metabolism using
Multi-agents System

• Metabolic pathways describe chains of enzymatic
  reactions.Their modelling is a key point to
  understand living systems. An enzymatic reaction
  is an interaction between one or several
  metabolites (substrates) and an enzyme (simple
  protein or enzymatic complex build of several
  subunits)
• Many ordinary differential equation models are
  available in the literature.They well fit
  experimental results on flux values inside the
  metabolic pathways, but many parameters are
  difficult to transcribe with suchmodels
  localization of enzymes, rules about the
  reactions scheduler, etcMoreover, a model of a
  significant part of mitochondrial metabolism
  could become very complex and contain more than
  50 equations.
• In this context, the multi-agents systems appear
  as an alternative to model the metabolic
  pathways.

16
Mitochondria
17
Potentials Deformations Applied to a Simple
Molecule
18
Simulation of Mitochondrial Metabolism using
Multi-agents System

• A abstracted 3D structure.
• B application to a phospholipid.

19
References

• 1 Definition of an agent by Prof. Michael
  Wooldridge.
• JADE Java Agent Development Framework.
  http//jade.tilab.com
• A-Globe A-Globe Agent Platform.
  http//agents.felk.cvut.cz/aglobe
• Jung-Jin Yang, An Ontology-Based Intelligent
  Agent System for Semantic Search in Medicine
• Antonio Moreno, Medical Applications of
  Multi-Agent Systems
• Charles Lales, Nicolas Parisey, Jean-Pierre Mazat
  and Marie Beurton-Aimar
• Mitochondria from http//universe-review.ca/