A look at “Natural Intelligence”

IMMM
Interactive Medical Mind Maps
Open Source working group of the IsfTeH, International
Society for Telemedicine

Etienne Saliez
GNUHealth Con, November 2017, Las Palmas
 Introduction
● Up to now many medical record softwares were
not mach more than narrative reports of what
did appen at some time in the past in some
specific context.
● Student can find much static knowledge on
internet, but have to learn a “medical
methodology” about how to use this large
amount of knowledge.
 Objectives
● Evolution to a “task oriented model”, to help to
solve patients health problems.
● Visual support of "Natural Intelligence". Focus
on the relations between medical concepts.
● Intended for the training of students.
● Mind synchronisation between the members of
care teams across Internet
 Node types in the graph
● Observations: just the observed facts without any
judgement.
● Health Issues: concerns requiring attention, as
abnormal findings and hypothesis, as well
diagnoses.
● Actions: decisions about what to do, either more
explorations or treatments.
● Links to medical knowledge, also presented as
graph navigations.
 Attributes
● Degree of belief:
● Intensity, importance:
● Probability:
● Precision, range of probabilities:
● Handling of both positive and negative assertions:
● Completeness of information: if possible which kinds of information should be
researched:
● Reliability of the sources:
● Current status of an Health Issues: Active, Chronic, History
● Expected utility of a recommendation.
● Costs: a multidimensional concept including risk, disagreement, delay before
conclusion, money, availability of the resource.
● etc...
 Attributes representation
● ( A ) Internally as numeric values, in order to allow valuable
pattern recognition including weighting.
● ( B ) Rounded to free speech style (as for example "not to be
excluded", "very probable", "obvious", etc...) in order to remain
compatible with the traditional way of human spoken
communication.
● ( C ) As visual graph, where the numeric values are converted to
color, shape, size, thickness, etc... Visual representations are
expected to provide easier and faster understanding of the
patient situation by the other members of the Care Team of the
patient. Several options to be experimented in function of the
user preferences.
 Navigation
● Click on a node and get the related topics. Again
and again.
● Start from a problem and move to related:
– Health Issues,
– Observations,
– Actions
– Anatomical domain,
– Related Medical Knowledge,
– Etc…..
 Training
● The student start from a few Observations.
● He has to try to build a graph with hypotheses and
recommendations about what should be done.
● The teacher review quickly the graphs made by
students and coordinates the discussions.
● The team try to improve the graph to a better and more
realistic vision of the case, paying particular attention
at probabilities, relative importance, degree of belief,
risks, …. namely attributes of nodes and relations.
 Telemedicine
● Graph can help to provide up to date common
understanding between the members a care
team across Internet.
● As a first step of migration from existing
medical records, graphs may be used as a kind
of “table of content”, where nodes can open
traditional documents.
 Seeking partners for a prototype
● Medical community: participation in
experimental development and better
representation of natural reasoning.
● Technical community: how to solve many
technical challenges on gitHub.
● http://www.chos-wg.eu/Projects/IMMM/main-IM
MM-Overview.html
● Contact: etienne@saliez.be
 MEDFLOSS

Database of medical Open Source Softwares and

promotion of more collaborations
See
http://www.chos-wg.eu/Projects/main-MEDFLOSS
.html
Open Source working group of the IsfTeH, International
Society for Telemedicine
Etienne Saliez
GNUHealth con, November 2017, Las Palmas
 Medfloss.org
● We assume here you are already familiar with
Medfloss.org
● A large inventory of several hundred Open
Source medical software projects.
● However collaborations remain relatively poor
between Open Source projects.
● Could any project as GNUHealth alone make
everything needed ?
 Objectives

●Promotion of the reuse of shared software

components.
● The potential benefits of the Open Source

philosophy is mainly based on know-how

exchanges and on sharing development efforts.
 Approaches
● Trying to install large software "packages" in
any new site remains difficult.
● The approach is here to share at least
selections of smaller "software components”.
 Requirements analysis
● The first question is to try to define what we
ideally need.
● Try to make a clear distinction between what is
of “common” interest and what are “additional”
adaptations. Some discussion material:
http://www.chos-wg.eu/Modules/Modules-Overv
iew.html
● The next question will be to find what we can
achive in the coming year.
 Technical objectives
● Review of integration architecture available in
the public domain.
● Allow collaborations between components in
different programming languages.
● Allow collaborations between different servers.
● Resilience to poor internet communications.
Non blocking communications.
● Easy maintenance.
 Considered approaches
● "Event based", as ReactiveX, http://reactivex.io/
● A component emit an “event”,
● The event contains a payload, data type identification,
data and “callback” request.
● Other modules acting as "Observers" can get the event,
process it and create new event(s).
● These new event can be observed back by the inial actor.
● Modules remains largely independent.
● Events are “non blocking” the sender.
 Medfloss as a graph ?
● Of course Medfloss has already excellent
search functions, but could graph technology
be useful to navigate in this large database of
OPEN Sources projects ?
● Seek the interrelated functionalities I need.
● Check the dependencies.
 Call for partners
● Call fot partners willing to try this perpective
and to provide evaluations.
● Existing medical software projects need only to
add interfaces with well defined messages and
according to http://reactivex.io/languages.html
● http://www.chos-wg.eu/Projects/main-MEDFLO
SS.html
● Contacts: etienne@saliez.be and
tkaropka@gmail.com