Multi-agent systems applied in Health Care

Multi-agent systems applied in health careDr. Antonio MorenoITAKA-Intelligent Tech. for Advanced Knowledge AcquisitionComputer Science and Mathematics Dep.Universitat Rovira i Virgili, Tarragonahttp://deim.urv.cat/~itaka

Outline of the talk

Rationale for applying agents in health careSome specific projects developed by the members of ITAKA

Web-based platform for providing home care servicesManagement of clinical guidelines

Research and development challengesFinal thoughts

http://deim.urv.cat/~itaka

Health Care problemsDistributed knowledge

E.g. different units of a hospitalCoordinated effort

E.g. receptionist, general and specialised doctors, nurses, tests personnel, ...

Complex problemsE.g. home care management

Great amount of informationE.g. medical information in the Web

MAS applied in Health Care

Summary of main motivationsMAS are inherently distributedAgents can coordinate their activities, while keeping their autonomy and local dataDynamic and flexible distributed problem solvingmechanismsUse of personalisation techniques

Example: national organ transplant coordination

Growing interest

Specialised workshops at AA00, ECAI02, ECAI04, IJCAI05, ECAI06, AAMAS08

AI-Communications special issues (2003, 2005)Book on Whitestein Series on Agent Technology (2007)

Int. Workshop on Health Care Applications of Intelligent Agents – February 2003

Book on Whitestein Series on Agent Technology (2003)

AI in Medicine special issue (2003)IEEE Intelligent Systems special issue (2006)

Outline of the talk




© K4Care, 2006

KnowledgeKnowledge--Based Based HomeCareHomeCare eServices eServices

ffor an Ageing Europeor an Ageing Europe

Project Presentation K4CARE Consortium

A Project funded by the European Community under the Sixth Framework Programme for Research and Technological Development

Contract no IST-026968

K4Care basic facts

March 2006 – March 2009 (3 years)Extended until September 2009

EC funding: 3.130.000 €

STREP project, 6th FP

Thematic area: Information Society Technologies (IST)

Coordinator: University Rovira i Virgili

13 Partners from 7 countries

K4Care project

The aim of the K4Care European project is to provide a Home Care model, as well as design and develop a prototype system, based on Webtechnology and intelligent agents, that provides the services defined in the model

Basic features: a) Actors are members of well defined organizations, with

different roles and allowed activitiesb) There is extensive domain knowledge to be considered c) Coordination of tasks in daily care

K4Care Model: Structure1 Nuclear Structure + n Accessory Services

...

THE K4CARE MODEL

HCNS

Actor Service

Action Procedure Data/Information

K4Care Model: Actors and Teams

K4Care Knowledge structures

EHCR: Electronic Health Care RecordAPO: Actor Profile OntologyCPO: Case Profile OntologyFIP: Formal Intervention PlanProceduresIIP: Individual Intervention Plan

DBs, Electronic Health Care Record

Data Base: with information about the K4Care actors as users of the K4Care Platform (e.g. contact information)EHCR: with the data about the Home-Care processes performed within the K4Care Platform

Medical documents stored in XML

K4Care Ontologies (I)

Actor Profile Ontology (APO)Types of actorsActions related to each rolePlatform servicesProceduresDocuments...

K4Care Ontologies (II)

Case Profile Ontology (CPO) DiseasesSyndromesSigns and symptomsSocial issuesAssessment testsInterventions...

K4Care FIPsFormal Intervention Plans (FIPs) are formal structures representing the health care procedures to assist patients suffering form particular ailments or diseasesFIPs are represented with the SDA* formalism

StatesDecisionsActions

The SDA* formalism is used to representK4CARE Service ProceduresK4CARE Formal Intervention PlansK4CARE Individual Intervention Plans

FIP for the management

of hypertension

Procedures

Formal specifications, in the SDA* language, of the way in which an administrative service(e.g. admit a new patient to the Home Care service) has to be implemented

Knowledge layer

Definition of an Individual Intervention Plan

Input: patient data (EHCR), result of comprehensive assessment, general K4Care knowledge structures (APO, CPO, FIPs)Output: Individual Intervention Plan to be applied on a patientProcess:

Select set of applicable FIPs (diseases, syndroms, symptoms)Merge FIPsAdapt the resulting SDA* structure to the individual characteristics of the patient

K4Care platform features Agent-based Web-accessible platform that provides a set of basic Home Care services

Definition of IIPsApply IIP to the patient

Relevant aspectsDeclarative and procedural knowledgeSeparation of the knowledge description from the software realization Interaction between agents and end-usersAgent-oriented execution of patient-centred plans

Transparency between knowledge and its use

Interaction between agents and users

Multi-agent system

1 Actor Agent for each user, permanently runningWhen the user logs in, a Gateway Agent is dynamically created

Two-way communication Web-servlet-GA-AAWhen an Actor Agent has to manage the execution of a procedure/IIP, it creates dynamically a SDA-executor Agent

Agent-based execution of IIPs (I)

Agent-based execution of IIPs (II)

Agent-based execution of IIPs (III)

Agent-based execution of IIPs (IV)

K4Care main conclusions

KnowledgeIndividual Intervention Plans allow practitioners to implement accurate and personalised sequencesof actions for a particular patient’s treatment

UseThe architecture allows implementing agent-based coordination methods between the actors relevant in Home Care, which adapt their behaviour dynamically depending on the knowledge available in the platform

Outline of the talk




AgentCities.NET

AgentCities: European project (finished in October 2003) that aimed to build a network of open agent-based systems that provided intelligent services associated to a city

Our problem: to allow citizens and visitors of a city to have easy, flexible and secure access to medical information

Features: HeCaSe

The user may request information about all the medical centres available in a particular cityIt is possible to book a visit to a doctor The user may access his Medical RecordThe medical data about patients is protected and only authorised people can access to it (ciphered transmissions)

Clinical Guidelines (CGs)

Indications or principles to assist health care practitioners with patient care decisions Applicable in diagnostic, therapeutic, or other clinical procedures for specific clinical circumstances

CGs: benefits

Consistent clinical practice, avoidance of errorsReutilisation and tailoringRapid dissemination of updates and changesConsideration of appropriate knowledge at appropriate timeUse of formal representation languages

CGs: barriers in daily use

Lack of awarenessLack of familiarityInertia of previous behaviours

No integration with standard practices

Lack of time or resources

Automatic management and enactment of guidelines

Features: HeCaSe2

Medical services have been includedIt is possible to coordinate complex tasks between doctors and services, e.g. booking different services following some constraints Doctors can take decisions taking into account all the CG’s related resultsRepository of guidelinesMedico-organisational ontology

Knowledge representation46

Separate the medical and organizational knowledge from its actual use, to improve flexibility and adaptability to different clinical settings

Two different ways of representing knowledge: Procedural: contained in CGs

know-how Declarative: external to CGs and managed through an application ontology

know-what

Medico-organizational ontology

Create an application ontology for:

a) Modelling the healthcare entities with their relations

b) Representing all medical terminology used by all partners

c) Assigning semantic categories to those medical concepts

47

Medical ontology: main classes

Main classes:a) Agent class (information about internal

organisation)Information about Departments, Practitioners, Medical centres, Patient

b) Medical_domain (medical terms)Diseases, Personal data and Procedures

c) Entity and Event (semantic types, UMLS)Term categories and activities

48

Available relations (i): Object properties49

Object property DescriptionbelongsTo Any instance of a class that belongs to another

hasAssociatedProcedure A medical concept has an associated procedure. It is used by doctors to simplify a search (from UMLS)

hasResponsible Establish the responsibility of any medical concept that has to be performed by a healthcare actor

hasSemanticType Functional property to specify the semantic type of a concept

isAssociatedProcedureOf Inverse of hasAssociatedProcedure

isComposedBy If an instance a ∈ A belongsTo b ∈ B then, b ∈ B isComposedBy a ∈ A.

isResponsibleOf Inverse of hasResponsible

Available relations (ii): Data type properties50

Data type property DescriptionhasCUI Value of the CUI (Code Unique Identifier)

hasDescription Concept definition provided by UMLS

hasResult Type of output of an element (action or data concept)

hasResultBoolean Sub-class of hasResult that sets a Boolean as output

hasResultInteger Sub-class of hasResult that sets an Integer as output

hasResultString Sub-class of hasResult that sets a String as output

hasResultEnumerate Sub-class of hasResult that sets an Enumerate as output

hasResultComplex Sub-class of hasResult that sets a complex element formed by one or more simple results (concepts) as output

hasTUI Type Unique Identifier (TUI) of the concept

Execution of CGs (i)52

Retrieving the CG and patient’s dataThe doctor gets the appropriate CG through the GA At the same time, the DRA receives (proactively) the patient’s data through the MRA

The DRA gets (iteratively) the next stage to follow (decision, enquiry or action)

Decisions: The DRA interprets the logical conditions included in a decisionCurrent values of the required variables are stored in the patients’ health record

Execution of CGs (ii)53

Actions: the DRA should contact with the responsible of performing the required action

If the DRA wants to know information about the biopsyactionThe DRA contacts with the OA that knows:

A biopsy is a diagnostic procedureA biopsy is performed in a surgery departmentA surgeon belongs to the surgery dept.

Finally, the DRA contacts with an available surgeon to perform the biopsy

Execution of CGs (iii)54

Enquiries: the DRA collects data about the required variables

The enquiries (usually) are findings or properties that are introduced by doctors or nursesThe DRA may require other specialists to get themThe OA gives the information about the appropriate agent to contact

Agents for Clinical Guideline enactment

Outline of the talk




Some research topics on the use of MAS in Health Care

Communication standardsMedical ontologiesSecurity mechanismsImplementation of agents in mobile devicesPersonalised access to information

Less social and professional reluctance to adopt agent technology

Legal issues

General research topics on MASService description, discovery, compositionStandard agent communication languages and protocolsNegotiation, coordination, cooperation techniquesAgent-Oriented Software EngineeringTrustHuman-agent interactionIntegration with legacy software...

Outline of the talk




Some general thoughts (I)It is difficult to work with doctors

Very busy, unaware of technical details, change requirements…However, they may end up being happy with a rather simple system (e.g. a well-organised DB, statistics for annual report)

It is difficult to sell “agents” to hospital computer units

Understanding, maintenance, …Information systems are hospital-wide, centralised

Some general thoughts (II)

Security is a matter of degree …Sometimes “real life” technical issues make it unsuitable to use agents

Use of previous prototypes or programming languages

The frontier between “agents” and “non-agents” seems to be difficult to define

Multi-agent systems applied in health careDr. Antonio MorenoITAKA-Intelligent Tech. for Advanced Knowledge AcquisitionComputer Science and Mathematics Dep.Universitat Rovira i Virgili, Tarragonahttp://deim.urv.cat/~itaka

Multi-agent systems applied in Health Care

Technology

k4care ontologies

health care procedures

k4care projectthe

home care model

daily care

k4care european project

home care managementgreat

homecare processes