of medical decision making 1 The CREDO stack John Fox Department of Engineering Science University of Oxford and OpenClinical Thanks to Psychologists InformaticsCSAI ID: 195513
Download Presentation The PPT/PDF document "A common ground theory" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
A common ground theory of medical decision-making 1: The CREDO stack
John Fox Department of Engineering Science University of Oxford and OpenClinicalSlide2
Thanks to …Psychologists, Informatics/CS/AI
Andrew CoulsonIoannis Chronakis
Subrata
Das
David GlasspoolOmar KhanPaul KrauseSimon ParsonsMor PelegAli RahmanzadehMatt SouthRory SteelePaul Taylor Richard Thomson
Clinicians
Alyssa
Alabassi
John Bury
Robert Dunlop
John
Emery
Marc
Gutenstein
Andrzej
Glowinski
Mike
O’Neil
Vicky Monaghan
Vivek
Patkar
Jean
-Louis Renaud-
Salis
Robert Walton
Matt Williams
Guy Wood-Gush Slide3
SummaryMedicine is a rich and challenging domain for decision science and decision engineeringIt raises major challenges and curiously neglected questions at many levelstheory, technology, applications and more …The
common ground theory aims to provide a general framework in which toPromote discussion across disciplinesClarify research questions and Develop practical solutions
The CREDO stack is a
particular
instance, but there are many othersSlide4
The borders of the common ground“Prescriptive” (axiomatic, rational) theoriesLindley “there is only one correct way to take a decision”EUT, Multicriteria DT, game theory, … and many ad hoc variants
“Descriptive” (empirical, explanatory) theoriesCognitive (Nobel Laureates - Herbert Simon, Daniel Kahneman)Neuroscience (neuroanatomy, neuropsychology, “hot cognition”)
Ecological (e.g. Gary Klein “naturalistic” theories
)
“Practical” (engineering, design) theoriesDecisions are often framed and made with respect to standard practiceDecision systems may need to engage with accepted practiceSlide5
Medical motivation:Quality and safety of patient care
UK National health serviceVincent data on medical error in Acute Hospitals>10% acute hospital admissions in NHS lead to avoidable medical error
US
Institute of Medicine
IOM: “To err is human”; “Crossing the quality chasm”McGlynn: Quality of Health Care Delivered to Adults in the USASlide6
Quality of Health in the USAMcGlynn NEJM 2003Slide7
Classical definition of DMDecision-making [is] a cognitive process resulting in the selection of a belief or a course of action from among several alternative possibilities.
http://en.wikipedia.org/wiki/Decision-makingSlide8
The CREDO stackSlide9
Diversity of medical decisions Screening for and classification of hazards; Risk stratification and management;Selection of
tests and investigations; Diagnosing the cause(s) of clinical complaints; Selecting treatments and other interventions; Prescribing drugs (routes, dosages, polypharmacy
etc.);
Referring patient to a colleague
Deciding whether a decision is needed;Initiating, adjusting and stopping treatments; Deciding whether earlier decisions are correct or not; if not why not; adjust; reverse, reframe, retake; Slide10
Diversity of medical decisions Screening for and classification of hazards; Risk stratification and management;Selection of
tests and investigations; Diagnosing the cause(s) of clinical complaints; Selecting treatments and other interventions; Prescribing drugs (routes, dosages, polypharmacy
etc.);
Referring patient to a colleague
Deciding whether a decision is needed;Initiating, adjusting and stopping treatments; Deciding whether earlier decisions are correct or not; if not why not; adjust; reverse, reframe, retake; Slide11
MDM is reason basedRefer to specialist colleague if …There is a possible life threatening condition
I don’t know what to do or lack sufficient knowledgeThe NICE clinical guideline says I shouldPatient is eligible for a research trial Difficult patient, and I can’t resolve issue by myself
Patient has asked to be referred
Colleague or mentor has suggested I should …Slide12
MDM is dynamicDecision-makers must deal with changing and often unpredictable circumstances Decisions are not just choices, they are points in an evolving narrative (patient and professional) Common ground theory should address the
whole cycle of decision-making: When is a decision needed? what is the goal of the decision? What knowledge and strategies are relevant?
When is it appropriate and safe
to commit?
When is it necessary to revisit and revise commitments as the situation evolves?Slide13
MDM is reflectivePeter Pritchard, a now retired GP (2004):I am committed to putting the patient firstI respect the patient’s identity, dignity, beliefs, and values
I am open to self-criticism and self-auditI am open to peer criticism and peer auditI try to provide care of high quality and apply evidence-based medicine where appropriateI am dedicated to lifelong reflection and learning.Slide14
Example: cancer careSlide15
Example: cancer careSlide16
Example: decisions in contextSlide17
The CREDO stackSlide18
Agent theoryCognitive agents engage with their environment in perceiving, acting and communicating (with the clinical team and patients). From these engagements, cognitive agents form and modify beliefs about a current situation, leading to goals that guide their behavior over time
. Cognitive agents draw upon substantial (sometimes prodigious) bodies of knowledge, both general and specialist (e.g. medicine) Other key cognitive functions include abilities to reason,
frame
and
make decisions, formulate plans and schedule tasks. All these processes are subject to uncertainty, requiring different kinds of cognitive control, including ‘reactive’ (situation-driven) and ‘deliberative’ (goal-driven). Slide19
Metacognition and decision-makingKey features of humans in general (and medical professionals in particular) are that we Autonomously recognise the need for decisions, frame them
and make them as circumstances evolve;Can reflect upon the rationales for our beliefs, goals, decisions and plans
C
an review and modify our commitments and intentions as circumstances change)
None of this is addressed in classical decision science or decision engineeringSlide20
Dynamic decision-making A synoptic view Detect a problemFrame decision (specify goal and decision type)Assemble options
which might resolve the problemIdentify relevant data and criteriaConstruct reasons for/against options Engage with uncertainty, values, preferences
Aggregate reasons to assess relative
merit
of optionsCommit to a decisionImplement the decision (actions, plans etc).Monitor outcomes against goals and respondCycleSlide21
A common ground theory
From decision science to decision engineering: the CREDO stack
ResearchGate
2014
Beliefs
Commitments
Plans
Goals
Options
ActionsSlide22
Options
Commitments
Beliefs
Plans
Goals
Actions
Example: risk assessment
Moderate
risk
Worried,
well
Population
or
moderate or
high
risk
Genetic, statistical &
other
lines of reasoning
Assess
riskSlide23
Example: test selection
tr
Options
Commitments
Beliefs
Plans
Goals
Actions
Pain,
nodule
Ultrasound
Mammogram
CT etc.
Age,
symptoms, …
Family history
Mammogram, ultrasound
Investigate for possible cancer
Order
Mammogram &
ultrasoundSlide24
Reasons and decisionsArgument construction Knowledge U
Data LA (Claim, Reason, Qualifier)
Argument aggregation
{(Claim, Reason, Qualifier)} Agg (Claim, Modality)
Fox et al
ECAI, 1992; UAI 1994; Fox and Das, 2000
Krause
et al
Computational Intelligence
1995Slide25
Uncertainty and arguments Quantitative [0,1] degree of belief (e.g. probability, possibility)[-1,+1] bipolar measures (e.g. belief functions){1,2,3,…n} ad hoc weighting of arguments
Qualitative + “supporting”
arguments
{+,-}
“supporting” and “opposing” arguments{++,--, +, -} … plus “confirming” and “excluding”Modal Linguistic (perhaps, possible, probable, plausible …)Slide26
Ten features of argumentation in decision-makingArgumentation is a process of constructing reasons for or against competing claims. The background
knowledge (theories) which is used in constructing an argument may be specific to a particular domain such as medicine or law, or embody general principles that are applicable in all domains.Arguments increase or reduce confidence
in a claim, though we may not be able to be precise about
its quantitative impact.
The more independent and valid lines of argument we may construct in support of a claim the greater the confidence that is warranted in the claim the more independent lines of argument against the greater the doubtIn some cases a single argument can be conclusive – it confirms or refutes a claim absolutely. Furthermore, one argument may appear to conclusively support a claim, while another conclusively supports a logically contradictory claim. Tolerance of contradictions makes sense because arguments can be based on different background assumptions; a formal treatment should be similarly tolerant.Arguments and theories can themselves be objects of reasoning e.g. “I do not accept your argument that my theory necessarily predicts climate change because you are making unreasonable assumptions about the physics of the greenhouse effect”.Slide27
Ten features of argumentation in decision-makingSome arguments may be stronger and take precedence over others, leading to the rebuttal of one argument by another
Similarly some arguments may corroborate or buttress others, thereby strengthening the claim.In the absence of information about relative strength contradictory arguments can still play an important part in analysing evidence and making decisions.
Natural language provides an expressive vocabulary for discussing evidence.
It
would be desirable to develop techniques which use sound formal and mathematical languages for argumentation tasks but which can be translated to and from intuitive, natural language forms.If a rational agent is forced to choose between two or more competing hypotheses or actions it should choose the one in which it has the greatest overall confidence that it is the most credible (hypothesis) or the most beneficial (action), unless there are grounds to suspect that the current order of preference is not to be relied upon.A rational agent that is not forced to choose may defer a decision on the grounds that the arguments are inconclusive,unreliable otherwise unwarranted Slide28
Formalising the common ground theorySlide29
Common ground theory (1): DM
Decision making
Planning
John Fox1,2*, Richard P. Cooper3 and David W. Glasspool4
A canonical theory of dynamic decision-makingFront. Psychol., 02 April 2013 Slide30
The CREDO stackSlide31
ConceptsSymbols
Descriptions
Rules
Decisions
PlansClass hierarchies, semantic networksDiseases, Symptoms, Findings, DrugsMedical facts, Clinical notes Alerts, reminders, interpretations
Reasons
(arguments,
evidence
,
preferences)
Care pathways, workflows
Terminologies, coding systems
The knowledge
ladder
Agents
Expert systems, Personal care agentsSlide32
The CREDO stackSlide33
Decision engineering(See wikipedia article “decision engineering”)… it is possible to design
decisions using proven engineering methods used for designing other “objects” like bridges, buildings …A shared language of standard components … readily understood by all stakeholdersSoftware tools for design, development and deployment of apps and agents
Populate generic decision models (
Dx
, Tx, Rx …) with domain-specific (medical) knowledgeSlide34
PROforma: Reification into “tasks”Fox et al, MIE 1996; Fox and Das, AI in hazardous applications
, MIT Press, 2000
Plan
Decision
Enquiries
Actions
Candidates
Commitments
Beliefs
Plans
Goals
ActionsSlide35
Decision engineeringSlide36
The CREDO stackSlide37
Applications
Care pathways in cardiology
UPMC (USA
), NHS (NZ) , NHS UK
Diagnosis and treatment in endocrine conditions (thyroid, diabetes)
AACE (USA)
Decision support for general practitioners
BPAC (NZ)
Triage for common conditions
NHS Choices (UK)
Supporting the breast MDT- Royal Free Hospital
BASO 2008, ASCO 2009, BMJ Open, 2012
Triple assessment of suspected breast cancer
Brit J Cancer
2006
Chemotherapy for children with acute lymphoblastic leukaemia
Brit J Haematology
2005
Planning care for women at risk of breast/ovarian cancer
Methods of Information in Medicine
2004
GP referrals for common cancers
MEDINFO 2003
Genotype of HIV+ patients interpretation and selection of anti-
retrovirals
(
Infer
Med
, Hoffman la Roche)
AIDS
2002
Genetic risk assessment
BMJ
1999, 2000
Support for mammographic screening
Medical Imaging
1999
Prescribing in general practice
BMJ
1997Slide38
The CREDO stackSlide39
Decision support: human interactionSlide40
Where next for decision science and engineering?Embedded cognitionTime, space, objectsPlanning and acting, safely Symbolic cognitionPerception LanguageLearning
Multi-agent collaborationSlide41
SummaryMedicine is a challenging domain forUnderstanding human error and expertiseDeveloping decision theory, empirical science and engineering methods
It raises many important questions and some strangely neglected onesThis will require contributions from many disciplines but there is a high level of fragmentation in decision scienceThe “domino” is a first draft of a common ground theory, to promote interdisciplinary discussionT
he
CREDO stack
validates the theory to a first approximation demonstrates its practical value