Adaptive%20Systems%20in%20Healthcare

1
Adaptive Systems in Healthcare

• Velma L. Payne
• October 5, 2005

2
Papers Reviewed

• Core paper Adaptive Systems in Healthcare
• A. Cawsey, F. Grasso, C. Paris
• The HOMEY Project a telemedicine service for
  hypertensive patients
• T. Giorgino, S. Quaglini, C. Rognoni, J.
  Baccheschi
• Dialogue Systems for Health Communication ,ed
  AAAI Press ,pages 32 - 35 ,(2004)
• http//www.csc.liv.ac.uk/floriana/UM05-eHealth/Gi
  orgino.pdf

3
Papers Reviewed

• Adaptive Medical Workflow Management for a
  Context-Dependent Home Healthcare Assistance
  Service
• L. Ardissono, A. Di Leva, G. Petrone, M. Segnan,
  M. Sonnessa
• Proceedings of the CWS-05 International Workshop
  on Context for Web Services
• http//www.di.unito.it/liliana/EC/CWS05.pdf

4
Agenda

• Goals of Patient Information
• Considerations for Adaptive Information Systems
• User Model Requirements
• Adaptation Techniques
• System Evaluation
• System Implementation
• Applications of Adaptive Healthcare
• Conclusion

5
Goals of Patient Information
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Inform Patient
• Of their condition
• Of treatments, drugs, side effects
• Enable Decision Making
• Persuade Patient to be Compliant
• To follow course of action (improve diet,
  exercise, quit smoking)
• Reduce Stress and Anxiety
• Enhance Patient Care

6
Considerations for Adaptive Information Systems
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Each patient has different
• Information Needs
• Emotional Needs / Stability
• Cognitive Ability - ability to comprehend
• Stress Loads

7
Considerations for Adaptive Information Systems
cont
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Patients must trust source of information
• Patients need to be in control
• They need to control the information the system
  keeps about them and their condition
• Systems must maintain patient privacy
• Systems must keep patient Information secure

8
User Model RequirementsWhat needs to be
captured in order to provide tailored medical
information to a patient?
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Patient Data age, sex, diagnosis, past and
  current treatments, specialists that have been
  seen, etc.
• Mental and Emotional State
• Acceptance of Their Disease
• Ability to Make Decisions
• Ability to Perform Complex Tasks
• Motivation Level / Willingness to Change
• How do you evaluate these ?

9
User Model Requirements cont
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Literacy Level
• Monitor patients state
• Update user model with state changes
• The author designates the last two items are
  research challenges and does not expound upon
  how a system should accomplish them

10
Adaptation Techniques
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Natural Language Generation (NLG)
• NLG techniques are concerned with the automatic
  production of appropriate textual documents from
  structured data sources
• Two Stages of NLG
• Content Planning - planning what to say
• Information Presentation - deciding how to
  express information

11
NLG - Content Planning
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Input
• Domain information from a database or knowledge
  base
• User Model Data
• Process
• Selects information based on user data data
  available in Database or Knowledge Base
• Output
• Sequence of primitive messages

12
NLG - Information Presentation
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Ask / Consider the Following
• Should each fact constitute one sentence or
  should they be conjoined?
• Should facts be announced bluntly or indirectly?
• Should facts be presented formally or informally?
• What words should be used?

13
Adaptation Techniques
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Autonomous Agents (Ardissono paper)
• Autonomous Agents are software entities that are
  capable of independent actions in dynamic,
  unpredictable environments
• Autonomous Agent research has developed advanced
  action execution techniques that support
  sophisticated selection of actions to perform in
  order to achieve goals, taking contextual
  conditions into account

14
Autonomous Agents
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Belief Desire Intention (BDI) Agents
• Deliberative agents acting in an environment to
  achieve goals in a context dependent manner
• Capabilities of a BDI Agent are called Actions
• BDI Agent has an Action Library specifying all
  the actions it can perform
• Each action is aimed at achieving a Goal
• Each goal is characterized by its Applicability
  Conditions
• Each action has a Body describing the sequence of
  steps to be executed

15
Autonomous Agents
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• A1 Book Blood Test Goal Get Blood Test
  Appointment

• Book Blood Test 1
• Goal Get blood test appointment
• Appl Condition Moveable patient
• Body
• Fix appointment with lab
• Notify patient about place
• and time of appointment

• Book Blood Test 2
• Goal Get blood test appointment
• Appl Condition Non-moveable patient
• Body
• Fix appointment with lab
• Request nurse at home
• Notify patient about the arrival
• of nurse
• Notify nurse about lab test

16
System Evaluation
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Need to consider
• What is the claim or hypotheses ?
• What techniques can be used to prove or disprove
  the validity of the hypotheses ?

17
System Evaluation cont
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Questionnaires
• Is system preferred over other systems
• Does system provide relevant information
• Increased patient understanding
• Reduction of anxiety
• Higher level of compliance

18
System Evaluation cont
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Randomized Clinical Trial (RCT)
• Patients are randomly assigned to an intervention
  (personalized / non-personalized system)
• Patients participate in the study
• Results are collected and measured statistically
• Conclusions are determined

19
System Evaluation cont
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• The author recognizes
• RCT is the gold standard for evaluation within
  the medical domain
• It is difficult for computer scientists to
  utilize this method of evaluation for information
  systems
• Other means of evaluation of information systems
  are needed for information system evaluation
• but the author did not provide any insight on
  this issue

20
System Implementation
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• How to implement an adaptive information system
  into the medical community was raised, but not
  answered, by the author

21
System Implementation cont My Thoughts
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Gain support from a entity in the medical
  profession that is open to the use of computer
  systems
• Integrate such an entity in the project from the
  get go
• Ensure the entity recognizes they are a valuable
  contributor to the project
• Ensure the entity recognizes the system will
  enhance their business and reduce costs
• Assist the entity in the introduction and
  tracking of the system to the patient

22
Applications - Piglet
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Concerned with creating personalized materials
  for patients undergoing treatment for diabetes
  and cancer
• Provided patients with on-line access to their
  medical record
• Provided hyperlinks to explanatory information
  about their condition and treatment
• These explanatory pages were generated
  dynamically from a knowledge base of medical
  information and patient record

23
Applications - Piglit
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion
24
Applications - Piglit
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Adaptive Techniques
• Natural Language Generation
• Evaluation
• Randomized Control Trial 525 patients (part
  with personalized info / part with general
  information system)
• Questionnaires patient views of system and
  anxiety level before and after intervention
• Results
• Patients receiving personalized information felt
  they learned relevant information

25
Applications HOMEY
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Personalized Home Monitoring to Support
  Continuity of Care
• System Goals
• Ensure chronic patients with hypertension and
  diabetes comply to the therapeutic and lifestyle
  regimes
• Improve information flow between patients and
  medical staff

26
Applications HOMEY
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Allows patients to communicate their test values
  to the medical staff
• System updates patients records
• System prompts / questions / makes
  recommendations to patients on their lifestyle
• Medical staff reads patients updates and
  provides recommendations

27
Applications HOMEY
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Adaptive Techniques / System Design
• Natural Language Dialog Technology / Intelligent
  Dialogue Technology - provides real time dialog
  rather than canned scripts
• Creates dynamic html pages and voiceXML sentence
  fragments
• System remembers past conversations / asks about
  specifics of these. For example, system could
  day You wanted to quit smoking, have you began
  to reduce your smoking habits yet?

28
Applications HOMEY
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Sample Dialogue
• System What is your heart rate?
• Patient What do you mean?
• System How many times a minute does your heart
  beat?
• Patient My heart rate is 67, my BP is 90 over
  120
• System Are you still swimming twice a week?
• Patient Yes
• System Are you still smoking?
• Patient Yes, 5 cigarettes per day
• System You should stop smoking

29
Applications HOMEY
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications
Conclusion

• Evaluation
• Study performed in two Italian hospitals
  hypertension units / Two different studies
• Randomized Control Trial 300 patients
• Blood Pressure Measurement before and after
  system use
• Results
• Patients using system had greater systolic
  pressure decrease

30
Applications Home Healthcare Assistance Service
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications
Conclusion

• Goals
• Allow patients to spend time in a familiar
  environment
• Efficiently manage home health-care assistance
• Reduce hospital expenses
• Methods
• Provide a virtual connection between patient and
  hospital
• Virtual connection monitors activities, requests
  services and manages medical protocols that the
  patient under-goes

31
Applications Home Healthcare Assistance Service
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications
Conclusion

• If multiple guidelines apply to a patient, they
  are carried out as parallel work-flows, each
  managed by the Guideline Manager
• The active guidelines have to be coordinated in
  order to prevent scheduling conflicts
• Updates to the patients clinical record are
  synchronized with patient data, pending
  appointments and tasks the patient should perform

32
Applications Home Healthcare Assistance Service
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications
Conclusion

• Authors did not provide any information on the
  evaluation of the system
• The paper only described the development of the
  framework

33
Conclusion
Goals Considerations User Model Techniques System
Evaluation System Implementation Applications Conc
lusion

• Adaptive systems definitely have a place in the
  Healthcare Industry
• The development of such systems can be complex
• Integrating such systems into the industry
  remains a challenge which must be overcome

34

• Thank You