Meeting starts: 9:05pm
Meeting Purpose
To determine if automation is possible for Stage I E$$ENTIAL MEDICINE$, Australia
Chair
Alice Motion (A) (@alintheopen )
Attendees (please add your name or reply below)
USYD: Hung Phat Duong @fantasy121 (Minutes), Yaela Golumbic (Y) @yaelago, Kymberley Scroggie (K) @kym834,
GitHub: Mitchell Blyth (M) @Ecaloota
Alice welcoming everyone to project and acknowledging the traditional custodians of the lands
Agenda
• Brief Project Update
A: Couldn’t work with high school students in normal manners due to COVID-19
leading to development of E$$ENTIAL MEDICINE$ (E$) project
2 stages so far, US focus
Stage I: understand US medicine availability -> automation of process. Prior manual work validated automation.
Citizens get to find solutions to contemporary problems without definite answers
Leads to Australian-side of E$, launched at National Science Week (NatSciWk)
What process can be automated and what cannot be.
• Discussion of Stage I Goals (EM Australia)
K What US drugs are available in Australia?
Similar workflow to what was done in stage 1
Y: Complementary to US E$ database
M:
Data sources
Australian therapeutic good register (as K mentioned)
A: Link to the website to search the ARTG
https://www.tga.gov.au/australian-register-therapeutic-goods
K: Accessible file size? Or just PDFs?
M: Reduce info to text to extract it
Y: Partial automation possibilities? Or some things just need to be manual?
K: share screen of ARTG website (similar process to Orange Book for US Stage 1)
Eg: of one drug abacavir
Details of search result
Incl PDF with additional info
Looking for ways to extract info from these results
Useful info in PDFs
Needs:
• ID name (generic and trade)
• ID numbers
• Strengths of dosage
• Years of approval
M: Consistency of formatting in PDFs?
K: Yes
M: Which version of approval is best interesting? (first or latest)
K: Needs more examples to know for sure
A: Best if collecting all info of approval
• Discussion of database requirements & Discussion of automation possibilities
A: How to meaningfully share these data? Best approaches to database for Australian E$.
M: Longevity. Permanent database up to 10 years or more.
Y: Share database (crowd-source citizen) is a bit more messy than fully automated info
A: When data can be published?
Info about change of ownership of medicines.
Automatically published data right away
Or
goes through multiple refinements before published
Or
like Wiki where info goes up right away but can be further refined
Y: Stage II more complex questions and in depth research into history of E$. Harder to find for some and info could be incomplete
How to structure database for this? (with some missing info) and not interrupting flow
M:
Query -> result (need sanitisation step before result is displayed)
OR
Website/Wiki then needs mechanism for editing
Example of visual representation (as prepared by K): good but takes too long.
Code info to output time code/visuals
M: Can automate data into visuals. Might be difficult with multiple drugs. What are the major events (Identify these)?
A: Convergent or divergent paths of each medicines to form therapies. Needs more examples or these. How frequent are these major events?
• AOB
N/A
• Next Steps and Action Items
ACTION ITEMS:
Kymberley: Other sources of info to data-mine from (Share on GitHub if anything turns up, then Mitchell will verify if automation is possible
Look through examples of compounds in Stage II for key events
Mitchell: Ways to automate data collection ARTG (Mitchell: also see if info can be extracted from PDFs) (Alice says timelines is a few weeks (2wks) to be in time for NatSciWk as more participants join. Mainly what can be done and what can’t be)
Ways to explore automation of data visualisation (Alice, but postpone for now until we know for sure what needs to be displayed).
(After database/few entries) Compare with US entries. Make connections, link together. (Compare time series/codes)
Looking for opportunities to publish different chapters of this research through blogs or journals
NatSciWk (18 Aug) and Hackathon (20 Aug), combo of stages I and II use this for what cannot be automated.
E_20200713_Minute_DUONG.docx
Meeting details here Meeting Recording here
Meeting starts: 9:05pm Meeting Purpose To determine if automation is possible for Stage I E$$ENTIAL MEDICINE$, Australia Chair Alice Motion (A) (@alintheopen ) Attendees (please add your name or reply below) USYD: Hung Phat Duong @fantasy121 (Minutes), Yaela Golumbic (Y) @yaelago, Kymberley Scroggie (K) @kym834, GitHub: Mitchell Blyth (M) @Ecaloota
Alice welcoming everyone to project and acknowledging the traditional custodians of the lands
Agenda • Brief Project Update A: Couldn’t work with high school students in normal manners due to COVID-19 leading to development of E$$ENTIAL MEDICINE$ (E$) project 2 stages so far, US focus Stage I: understand US medicine availability -> automation of process. Prior manual work validated automation. Citizens get to find solutions to contemporary problems without definite answers Leads to Australian-side of E$, launched at National Science Week (NatSciWk) What process can be automated and what cannot be.
• Discussion of Stage I Goals (EM Australia)
K What US drugs are available in Australia? Similar workflow to what was done in stage 1 Y: Complementary to US E$ database M: Data sources Australian therapeutic good register (as K mentioned) A: Link to the website to search the ARTG https://www.tga.gov.au/australian-register-therapeutic-goods K: Accessible file size? Or just PDFs? M: Reduce info to text to extract it Y: Partial automation possibilities? Or some things just need to be manual?
K: share screen of ARTG website (similar process to Orange Book for US Stage 1) Eg: of one drug abacavir Details of search result Incl PDF with additional info Looking for ways to extract info from these results Useful info in PDFs Needs: • ID name (generic and trade) • ID numbers • Strengths of dosage • Years of approval
M: Consistency of formatting in PDFs? K: Yes
M: Which version of approval is best interesting? (first or latest) K: Needs more examples to know for sure A: Best if collecting all info of approval
• Discussion of database requirements & Discussion of automation possibilities A: How to meaningfully share these data? Best approaches to database for Australian E$. M: Longevity. Permanent database up to 10 years or more. Y: Share database (crowd-source citizen) is a bit more messy than fully automated info A: When data can be published? Info about change of ownership of medicines.
Automatically published data right away Or goes through multiple refinements before published Or like Wiki where info goes up right away but can be further refined
Y: Stage II more complex questions and in depth research into history of E$. Harder to find for some and info could be incomplete How to structure database for this? (with some missing info) and not interrupting flow
M: Query -> result (need sanitisation step before result is displayed) OR Website/Wiki then needs mechanism for editing
Example of visual representation (as prepared by K): good but takes too long. Code info to output time code/visuals M: Can automate data into visuals. Might be difficult with multiple drugs. What are the major events (Identify these)? A: Convergent or divergent paths of each medicines to form therapies. Needs more examples or these. How frequent are these major events?
• AOB N/A
• Next Steps and Action Items
ACTION ITEMS: Kymberley: Other sources of info to data-mine from (Share on GitHub if anything turns up, then Mitchell will verify if automation is possible Look through examples of compounds in Stage II for key events Mitchell: Ways to automate data collection ARTG (Mitchell: also see if info can be extracted from PDFs) (Alice says timelines is a few weeks (2wks) to be in time for NatSciWk as more participants join. Mainly what can be done and what can’t be) Ways to explore automation of data visualisation (Alice, but postpone for now until we know for sure what needs to be displayed). (After database/few entries) Compare with US entries. Make connections, link together. (Compare time series/codes) Looking for opportunities to publish different chapters of this research through blogs or journals NatSciWk (18 Aug) and Hackathon (20 Aug), combo of stages I and II use this for what cannot be automated.
NEXT MEETINGS TBA
Meeting adjourns 9:46pm