medical collaborative expert (demo)
DESCRIPTION
TRANSCRIPT
Demo (2010):
Medical Collaborative Expert™
professionalsand consumersArbitrated Health Decisions for
working name
Gil Ronen © 2010
Founder/Demo Developer
http://www.linkedin.com/in/gilronen
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Our goals
Medical Collaborative Expert was a startup founded in April 2009 by Gil Ronen and Miki Kolko. Our goals were to (1)
promote participatory and shared decision-making between providers and patients by providing concise analysis of
medical information, (2) alleviate expertise bottlenecks by bridging the divide between lay users, prosumers and medical
professionals, (3) provide rigorous medical reputation & trust mechanisms that help discern quality and locate expertise.
The startup was called variously MyPulse and SecondOpinion. For more background see main MCE slide deck (this is
the demo slide deck).
?we were buildingWhat
A service providing personalized step-by-step instructions relating to medical and preventive-health decisions to
consumers, patients and medical professionals; what to do next with regards to treatments, self-tests and lab
procedures, monitoring and follow-up and how to go about them. The service utilizes a new collaboration platform that
addresses quality issues by incorporating proven social arbitration mechanisms and proprietary reputation and trust
algorithms and processes. As a by-product of the service a comprehensive evaluation network is created for content
creators providing help in locating expertise in specific topics and opportunities for professional recognition.
Progressively, the service would have supported multiple delivery methods and cater to varying levels of medical
knowledge with a focus on exploiting the continuum of knowledge from lay users, to prosumers, and to various degrees
of medical specialization.
Where is it now?
Between December 2009 and May 2010 Gil Ronen designed and developed a demo system using Python, Django and
MySQL. The startup was terminated in August 2010 having failed to secure funding alongside well-funded competition.
Synopsis
Mass collaboration, a proven technique,
utilized to generate Q&A sessions on
medical topics. Proprietary accreditation
and reputation algorithms and other social
structures arbitrate between conflicting
elements for state-of-the-art conclusions.
Knowledge authors share ad revenue
based on their effectiveness in high-quality
content production
Domain-specific searches identify more
materials, portals provide more
information, tools targeting professionals
identify articles and ongoing clinical trials
and provide less than cutting-edge
diagnosis,
Envisions personalized step-by-step
instructions relating to medical and
preventive-health decisions for consumers,
patients and medical professionals. Our
solution helps create quality knowledge
and identify where expertise lay
Our Solution Status Quo Solutions
Professionalsand ConsumersVision: Social Arbitration for
Why knowledge and not content?
Current analysis process: search the web ,
look for similar conditions, analyze articles ,
generalize from anecdotal cases on blogs,
filter biased content, find inconclusive and
contradictory conclusions, get confused,
bring stack of printouts to physician office
visit, request various tests and medications,
experience side-effects, start over
How do we create it?
Content Arbitrated Knowledge
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Demo
For further background see the Medical Collaborative Expert slides (not the demo).
Demo written in Python with Django and MySQL from December 2009 to May 2010.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Part I: Use Guide to Get Medical Recommendations
Search for a Guide and Select to Start
Visitors search for medical guides
to particular topics. A selected guide
can be invoked to provide
individually tailored
recommendations.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Answer Guide Questions and View Recommendations
A new guide may be invoked if the
user requires assistance in
answering a particular question.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Recommendations (actions).
Users answer guide questions to get
tailored recommendations. Each
additional response changes the
recommendations shown to the users..
Guide questions
Possible responses
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Part II: Author Guides Collaboratively
Login includes
locality for guest
users.
Entry point for
Collaborative Medical
Expert (aka
MyPulse.com,
)SecondOpinion.com
Log In as Registered User or Guest
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
The guides in the repository can be
applied en masse to a user’s online
medical records providing
recommendations that change over time
as guides are updated by the community
or new test results are entered into the
user’s medical record.
gilnyc logged in and
has a reputation score
of 97.4.
Dashboard for User ‘gilnyc’
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Find Guide is the main
screen for finding and
applying medical
guides.
Search for topics
related to ‘nursing
problem’.
A single medical
expert (guide) is
found when using
‘nursing problem’ as
the search keyword.
The guide can now
be either started or
edited.
Find Guide to Edit
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Guides are built collaboratively by
many users over time.
Edit ‘Nursing Problems’ Guide
A guide consists of questions that
specialize the recommendations
based on the user’s responses.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
View ‘Nursing Problems’ Guide Recommendations
Recommendations (Actions)
correspond to a pattern of
responses within the guide – when
a user starts a guide and responds
to questions, the
recommendations that closest
match the pattern of responses
rise to the top of the displayed
recommendations.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Edit Recommendation (Action) Triggers for ‘Nursing Problems’ Guide
A recommendation is associated
with a set of responses to the
guide’s questions. A question can
be skipped if its response is not
relevant to the recommendation.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Edit Supports for ‘See an Osteopath’ Recommendation
A recommendation is supported by
links to support materials such as
articles or documented anecdotal
evidence.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Phonetic Search
Phonetic search on ‘slip apiya’
results in two guides being
presented.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Logout user ‘gilnyc’ and login as
‘DrBob’. Notice lower reputation
score of 26.5.
Edit ‘Obstructive Sleep Apnea’ Guide
User DrBob with low reputation
score wishes to edit the question
‘Are you sleepy during the day’?
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Edit Question ‘Are you sleepy during the day?’
The question belongs to the
‘Obstructive Sleep Apnea’ guide
and consists of 3 possible
responses.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
The question also has three
recommendations (actions), two of
which suggest performing additional
tests.
Adding New Possible Answer
Adding a new possible answer (‘In
the afternoon’) to the question.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Position of New Answer
Answer is added as last response
option.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Part III: Collaborative Process
Proposed Deletion of Response Option
Dr Bob tries to delete an answer
option for one of the guide
questions but because of a low
user reputation score the
proposal requires additional
support from other users. In
essence, DrBob needs to pool
reputation with other users to
affect the proposed change and
make it available to others. Dr
Bob can vote on the proposal and
provide support and the proposal
will be pending other votes.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Crux of Collaboration: Providing Support for Content Proposal
Based on Argumentation Theory, supporting
an argument, or objecting to an argument, fall
under one of several pre-defined categories.
The user votes not only by applying their
reputation score to their support or objection
but are required to do so within a selected
category. Users are encouraged, but not
required, to provide additional materials for
the benefit of other prospective voters on the
proposal at hand.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Dr Bob selects to support his
proposal by choosing the ‘expert’
support category: DrBob wrote ‘the
book’ on Sleep Apnea and claims
to be an expert in the field. Of
course, citations and links are
needed to support this claim.
Recording Support for Proposal
Dr Bob’s vote is counted and the
current support and threshold for
acceptance of the proposal are
displayed:
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
An Arbitration Action is Required on the ‘Obstructive Sleep Apnea’ Guide
A different user logs in. Despite the
name, user ‘baba’ has a high
reputation score and therefore will
have more impact on proposals
than DrBob.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
The user selects to work on the
guide ‘Obstructive Sleep Apnea’
and sees that there is a question
that has a proposal pending. In the
parlance of the Collaborative
Medical Expert, it requires Social
Arbitration.
The user may have been
subscribed to this guide and
therefore received a notification that
arbitration is needed.
Social Arbitration of Proposal
User votes in support of proposal
using the Standard Procedure
support category.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Clicking on ‘what?’ for a category
will bring up an explanation, in this
case for the Research support
category.
Proposal Accepted
Vote is sufficient to cross the
Acceptance Threshold and so the
proposal to delete a response option
from the question is accepted.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
The same process is followed for other changes
to a guide. Depending on the Acceptance
Threshold of the proposal, the reputation of the
content and the reputation of the users involved,
the process may be immediate or more involved:
simple changes to low reputation content by
high reputation users would go into affect
without additional arbitration. For other content
it may be prudent to have a more involved
process of deliberation between knowledgeable
users through the Argumentation mechanisms
provided.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Part IV: Publish Guide Changes
Publish Guide on Obstructive Sleep Apnea
Click Publish to make guide changes
visible to guide users.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Part V: User Reputation and Content Reputation
User Reputation Enhancement
Logging in again as user ‘DrBob’
we see that the reputation score
was enhanced by acceptance of
the proposal authored by this user.
The increase is small reflecting the
value of the change. Note that the
user reputation score is not simply
cumulative but is in relation to the
scores of other users.
Gil Ronen © 2010 [email protected] http://www.linkedin.com/in/gilronen
Reputation mechanism highlights:
1. A user whose proposal is accepted sees an
increase in reputation.
2. A guide that is created by high reputation
authors will itself have a high reputation that
manifests as higher Acceptance Threshold
values for future proposed changes.
3. Positive ratings for a guide will reflect on
the reputation of its authors in correlation to
their contribution.
Appendix: Collaborative Medical Expert Server Log Snapshot