'LIVE' demo of a new social search tool, Coremine Medical, by PubGene AS. Coremine Medical is a powerful and flexible biomedical search tool. Using Coremine Medical you find relevant connections between medical concepts such as diseases, drugs, procedures, symptoms, genes, and medical experts. The search tool provides easy visualization and navigation and help save time, explore, comment and share. Coremine Medical maintain strong bonds to Oslo University Hospital (OUH) for research and relevant use cases, e.g. The Intervention Centre and integration of clinical data warehouse.

1. Coremine Medical
www.coremine.com
Build Your Biomedical Mind-Map
by
Dag Are S. Hov (PubGene AS)
Karl Øyri (OUS Intervention Centre)
Oslo, Norway

2. PubGene Background
• PubGene was founded in 2001
• Founders from NRH - Norwegian Radium Hospital and NTNU -
Norwegian University of Science and Technology (Jenssen, TK et
al., Nat. Genet., 2001)
• Original problem: Hjelp structure huge data sets in breast cancer
microarray research
• Pioneers in the use of text-mining and NLP to index Medline records
linking genes and diseases
• US patent for search algorithms and core technology in 2007

3. Coremine Medical
• Powerful and flexible biomedical search tool
• Finds relevant connections between medical concepts such as
diseases, drugs, procedures, symptoms, genes, and medical experts
• Easy visualization and navigation
• Help save time, explore, comment and share
• Strong bonds to Oslo University Hospital (OUH) for research and
relevant use cases, e.g. The Intervention Centre and integration of
clinical data warehouse

4. Case study
• You are a surgical resident taking part in a
group of colleagues working with the subject
‘pancreatic diseases’.

5. Goal for the day
• You want to find, explore and share
information using the advantages of Coremine
Medical.
• Also, you would like to search through
relevant electronic patient journals from the
integrated clinical data warehouse of your
hospital.

7. Enter the name of the
disease ’Pancreatic
Diseases’ and pick from the
suggestions provided by
Coremine.

8. 1) Continue to build your
input list,
2) deselect any categories you
are not interested in or
3) click ’Explore’ to create
your result network.

10. •Categories of concepts
related to your input
through PubMed records
are gathered in stacks
•Click a stack to view
related concepts from
categories like anatomy….

11. …MeSH (Medical Subject
Headings)...

12. …procedure...

13. …and drug.

14. •Bookmark the disease
concept into ’My Page’ to
keep it for later
•Login needed

15. •Not a user yet?
•A quick register will suffice
(no charge)
•Staying logged in will
enable complete search
history

17. utGTbHggwN
•Bookmark set after
registration completed
•All bookmarks are stored
under My Page >
Bookmarks

18. Logging in also enables
setting the email alert on
new articles

19. All ten synonyms are used
when linking the disease to
articles, as is the case with
all concepts in Coremine.

20. General info on the disease
is fetched from the U.S.
National Library of
Medicine.

21. Hierarchy info provides
hyperlinks to generate
networks on similar
diseases.

22. ’Extracted knowledge’
section shows you more
complete lists of the
concepts related to you
query.

23. ’Biomedical experts’
contains clickable names of
PubMed authors most
often appearing in articles
where Coremine finds
’Pancreatic Diseases’
(including synonyms).

24. ’Drugs’ shows related
drugs and the ’Load more’
link opens a continous
more complete list of drugs
in a falling degree of
relevance.

25. •The related concepts from
’Extracted knowledge’ can
be added to the network.
•Interact by mouse-click

26. Read the general drug info
and click ’Add to network’
to mix the drug with the
disease network on the
left.

28. •You may add to the
network as many related
concepts as you want.
•Here, concepts from
procedure and anatomy
are also added.

29. After spending time on
modifying the
network, you may want to
save it with its excact
layout.

30. You find the saved network
under My Page >
Bookmarks

31. •You can also click a
connection to focus on two
concepts.
•The right part of the
screen now displays info
relevant to both concepts

32. To create a network where
the added concepts are
used as normal input along
with the disease, click
’Click to modify your
search’.

33. If nothing more needs
tweaking, hit ’Explore’.

34. The related concepts in the
network are now
important to all four input
items.
•The network has now
become something you
might want to spread to
your friends or colleagues.
•Using the ’Share’ button
on the upper right, you can
easily share using email or
social media.

35. Selecting Twitter, the URL
to the result network is
sent to your followers.

36. Your followers can click the
Twitter link to create the
same network you just
viewed.

37. Clicking the Twitter link will
open a new window/tab in
a browser with the same
result network you just
shared.

38. •In your own result
view, you can use the
session history below the
network to get back to a
previous search result.
•Let’s go back to
’Pancreatic Diseases’

39. •Below the related
concepts, you find ’Related
articles and hits’ where
Coremine gets information
from external sources.
•Of the more authorative
sources, you first find the
recent PubMed articles

40. •You can bookmark any of
the articles.
•As for the bookmarks you
set for the disease and the
network earlier, the ones
for documents also end up
in ’My Page’.

41. •Another authorative
source is Clinical Trials

42. •The heading of any hit
under ’Related articles and
hits’ is hyperlinked to the
source.
•Bookmark any trial you
want to get back to.

43. •Bookmark any hit from
any supported source, like
Wikipedia, Google Image
and YouTube…

49. Share the bookmarks you
have gathered from My
Page to the Coremine
group ’Pancreas
group’, where you are a
member.

50. Bookmarks you share to
the Pancreas group can be
found and launched by the
other group members.

51. Another way you can
gather and share info from
result pages into groups is
using Context Notes under
’User updated content’.

52. •Your note will stick to the
disease in focus.
•Visability set to only a
group of users will ensure
privacy and notify the
other group members.

53. •Inside the group, you see
the note under the
’Context notes’ tab.
•Other members can reply
to the note inside the
group and in the Explorer
result page.

54. Under ’Group
management’ you can
invite others to the group
by Coremine nick name or
any email address.

55. For any hit under ’Related
articles and hits’ you can
also leave a comment.

56. •The comment will follow
the hit/document.
•Visibility can be set to the
same group as earlier.

57. The note on the document
is then also seen inside the
group ’Context Notes’ tab.

58. Specializing in Pancreatectomy
• Using the Coremine Library tool you want to
quickly get an overview of a large set of
documents by clustering them (feature in
development).
• You will also like to find focused information
on an expert in the field.

59. 2
1
3
To perform the clustering inside the Library
tool in Coremine, you start by searching for
the procedure concept ’pancreatectomy’.

60. •In this test example, 1000
articles has been selected
from the search result.
•You hit ’Cluster’ to start
the clustering of the
documents.

61. The algorithm needs a
number of clusters to
divide the documents
into, and a name for the
run.

62. •The result shows a quick
overview of what articles
was put into the same
cluster and what sentences
and keywords are the most
important.
•A better formatted
overview is found in
Explorer by hitting
’Network’.

63. You save the time to read the
articles to find what is the most
important content for you:
•Key sentences
•The top ranked
sentences, based on the
words inside the sentences.
•Keywords
•The top five most important
•Cluster articles (121)
•The cluster articles can be
bookmarked and
commented.
•Extracted knowledge
•Coremine supported
concepts found in the cluster
articles sorted in categories.

64. •Returning to the network of ’Pancreatic
Diseases’, you open the procedure stack and
click ’pancreatectomy’.
•As the focus is now on pancreatectomy, the
content of the section at the right changes
accordingly.
•You open ’Biomedical experts’ and click the
highest ranked expert, ’Büchler MW’ and
further follow the link ’Open as a new search’

65. •The author network shows
PubMed authors working with
similar subjects as Büchler MW.
•Basis for the similarity is
overlapping keyword profile.
•Each profile is built looking at all
the concepts found in an author’s
articles and how specific these
concepts are for these articles
compared with all other PubMed
articles.
•Green connections indicate co-
authorship

66. •For this expert on
’pancreatectomy’ you get
•Keyword profile – the basis
for comparing this author
with others

67. •For this expert on
’pancreatectomy’ you get
•Recent articles – read and
bookmark highlights.

68. •For this expert on
’pancreatectomy’ you get
•Publishing timeline – when
was this author productive?

69. •For this expert on
’pancreatectomy’ you get
•Biomedical experts working
with similar subjects –
complete list of the authors
best matching Büchler’s
keyword profile.

70. •For this expert on
’pancreatectomy’ you get
•Extracted knowledge – the
keyword profile sorted in the
Coremine categories.

71. Return to the network for
’Pancreatic Diseases’ using the
session search history below the
network.

72. Enterprise Clinical Warehouse Support
• In the following example we show how
internal enterprise data can be integrated.
• You reopened the network of ’pancreatic
diseases’ from the session history, opened the
anatomy stack and clicked the concept
’pancreas’ to focus and search for it in
electronic patient journals.

73. Focus is now on ’Pancreas’

74. •After scrolling down, you open
the section for the hits into patient
journals (electronic medical
records).
•This section will only be visible
inside the specific hospital.

75. •All patient journals where the
phrase ’pancreas’ is mentioned are
shown.
•You can bookmark journals.
•You enter a journal by clicking the
heading.

76. Opening the electronic journal you
can read about the patient and, in
this example, further follow a link
to a surgical procedure where the
patient had a laparoscopic
pancreatectomy performed due to
cancer.

77. •The page opened shows a unique
presentation from separate global
hospital applications in one context
based screen.
•By this you get a unique overview of
information associated to the particular
case.
•In the laparoscopic pancreatic surgery
example the main phases, episodes and
adverse events of the procedure are
tagged. The tags are displayed as
highlights of the surgery, so you get an
easy overview of the procedure that can
last for several hours.
•The surgery summary text is also
indexed, so you can quickly perform a
search for another patient journal or
create a Coremine network.
•You also get ’macro data’ like vital sign
data from the computerized order entry
and charting system and images from the
PACS system.

78. *
•’Micro/atomic data’ are also included exemplified by molecular biology data containing gene sequencing and genome analysis of
tumor cells. The molecular data in this example is a result of a pipeline analyzing the output from genome studies on the tumor
cells, like the output from 23andme, decodeme etc. The molecular data can be combined with other sources of information (eg.
drugs) and provide decision support for individualized chemo therapeutic drugs.
•This bridges the gap between clinic medicine and genomics in a novel way. The case shown here (figure on what tumor cell genes
are gained or lost in copy number on chromosome 17) is from real data where the patient got conventional chemo therapy that did
not cover the actual oncogenes eventually killing the patient (* Hjortland et al., BMC Cancer, In Press).

79. www.coremine.com - Summary
• Sophisticated search application in the public
domain facilitating immediate gathering and
intuitive presentation of multimedia-based
information from substantial publicly available
resources.
• Coremine can be integrated with complex clinical
enterprise data. The synergy facilitates tailored
individualized therapy based on phenotype in a
novel way. This contributes to bridge the gap
between applied clinical medicine and genomics.