II-SDV 2014 Design and development of a novel Patent Alerting Service (Bayer HealthCare, Germany)

742 views
571 views

Published on

Published in: Software, Technology
0 Comments
1 Like
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
742
On SlideShare
0
From Embeds
0
Number of Embeds
255
Actions
Shares
0
Downloads
22
Comments
0
Likes
1
Embeds 0
No embeds

No notes for slide

II-SDV 2014 Design and development of a novel Patent Alerting Service (Bayer HealthCare, Germany)

  1. 1. Design and development of a novel Patent Alerting System 2014-04-14 Dr. Wolfgang Thielemann
  2. 2. Slide No. 2 • 2014-04-14 Wolfgang Thielemann2 • Dr. Ortrud Steinführ • 23-JUN-2011 Agenda/ Content Introduction Workflow Using the platform to Search, Browse and Filter Email push service Summary
  3. 3. Slide No. 3 • 2014-04-14 Wolfgang Thielemann Introduction: What does this novel service do and why did we need it? 1
  4. 4. Slide No. 4 • 2014-04-14 Wolfgang Thielemann Patent Alerting System What does it do?
  5. 5. Slide No. 5 • 2014-04-14 Wolfgang Thielemann Patent alerting options existing at the start of project: • End-Users browse results of commercial alerting services like “Current Patents Gazette” • End-Users set up alerts in databases (e.g. Scifinder) themselves • Information professionals create alerts in various added-value databases and/or patent full-text databases Challenges for a novel, proprietary alerting service: • The number of newly published healthcare & chemistry patents is huge (2000-5000/week) • The bandwidth of topics to be tracked and the corresponding terminology is huge too A powerful, precise and focused alerting system is needed Patent alerting options
  6. 6. Slide No. 6 • 2014-04-14 Wolfgang Thielemann6 Using advanced text mining workflows, new patents have to be categorized into clearly defined, project specific folders Criteria for categorization should include: • Drug action / molecular target (PDE 5 inhibitors) • Specific medical condition (e.g. pulmonary hypertension) • Specific technology (e.g. positron emission tomography) • Compound class (e.g. antibodies) A The platform must be easy to use and end-user searchableC Minimal costs for creation, maintenance, uploads, license fees and hardwareD What did we expect from our new patent alerting system? The platform has to provide chemical structures which are representative for novel chemical space covered in medicinal chemistry patents B
  7. 7. Slide No. 7 • 2014-04-14 Wolfgang Thielemann Workflow 2
  8. 8. Slide No. 8 • 2014-04-14 Wolfgang Thielemann 1. Download of patent full-text 2. Filtering and categorization 3. Adding key content Overall workflow of novel patent alerting service Email push- service; RSS feeds Searching or Browsing APIs other prop. platforms Enrichment with chemical structures Patent full-text source Orbit Broad healthcare & chemistry related search for all alerts in: Chemical structure DBs CAS / Registry WPIX / DCR SureChem patent numbers of substructure hits for selected alerts Patent alerting platform Tabular patent sheets enriched & categorized with: • Project name • Indication • Molecular targets • Chemical structures • Technologies • …
  9. 9. Slide No. 9 • 2014-04-14 Wolfgang Thielemann The engine within the novel service 1. Download of patent full-text 2. Filtering and categorization 3. Adding key content 1. Download of patent full-text 2. Filtering and categorization 3. Adding key content
  10. 10. Slide No. 10 • 2014-04-14 Wolfgang Thielemann It’s not a commercial black box! We have full control over: • The process • The vocabulary • The rules … and can adjust it to the needs of our organization! The engine within the novel service
  11. 11. Slide No. 11 • 2014-04-14 Wolfgang Thielemann 2000-5000 newly published healthcare and chemistry related patent applications per week (first published member of a patent family + first US or EP application) Details of proprietary categorization, filtering, indexing steps + + + In-depth text mining analysis of the patent full-text to extract, standardize and add key terms (targets, indications, technologies etc.) Adding key content typically 0 - 5 patents per alert / week In-depth text mining analysis of the patent full-text for identification of relevant patents Filtering & Categorization
  12. 12. Slide No. 12 • 2014-04-14 Wolfgang Thielemann Added key content • Alert relevant keywords (e.g. drug action) • Indication • Molecular target (official NCBI Gene name + Gene ID) • Formulation (Route of Admin + Dosage forms) • Species • Technologies (e.g. prodrug, freeze drying, pegylation etc.; can be augmented to the needs of the organization) • Molecule type (small molecules, biologicals, natural products) • Patent type (compound, formulation, method general, diagnosis, preparation method, combination etc.) Keywords relating to the following topics are extracted, standardized and added: + + + + + + + +
  13. 13. Slide No. 13 • 2014-04-14 Wolfgang Thielemann Details of enrichment with chemical structures identifies chemical compounds: • from names (incl. IUPAC, brand names, generic names, trivial names) within 24 h after publication of a patent (WO, US, EP) • from drawn structures (only high quality structures without variables) within 2-3 days after publication of a patent (WO, US, EP) We add these structures to the new patents within the patent alerting workflow * *will soon become SureChEMBL
  14. 14. Slide No. 14 • 2014-04-14 Wolfgang Thielemann Tuesday Wednesday Thursday Friday Saturday Sunday Monday New WO New US New EP Name2str Name2str Name2str Conversion chem drawings Conversion chem drawings Conversion chem drawings Generation chemical structures Timelines: Providing alerts as fast as possible • Download • Categorization • Text Mining • Upload Alert …+ documents from other patent offices
  15. 15. Slide No. 15 • 2014-04-14 Wolfgang Thielemann Inclusion / Exclusion criteria The keywords are mentioned: • 1x in core fields (e.g. title) • Multiple times in other fields (e.g. description) The keywords are mentioned: • a few times in a non-core field Documents & keywords Chemical structures • Novel compounds (incl. intermediates) with a global frequency of <= 10 in all patents* which also pass our chemical purging filter • Common reagents, catalysts, or drugs which are often mentioned in “washing lists” * General WO, US, EP patent database Selection criteria for novel patents
  16. 16. Slide No. 16 • 2014-04-14 Wolfgang Thielemann Using the platform to Search, Browse and Filter 3
  17. 17. Slide No. 17 • 2014-04-14 Wolfgang Thielemann Alerting System Main Navigation: Subscriptions & Search Shows alerts you have subscribed to All other available alerts Google like search (keywords or chemical structures) with facetted filter options Administration of alerts (only for information professionals)
  18. 18. Slide No. 18 • 2014-04-14 Wolfgang Thielemann Alerting System Main Navigation: Entry page: “My Subscriptions” Information about scope of the alert Stop subscribing to the alert Show list of all documents collected for this alert so far
  19. 19. Slide No. 19 • 2014-04-14 Wolfgang Thielemann Alerting System Main Navigation: Other available alerts Information about scope of the alert Subscribe to the alert Show list of all documents collected for this alert so far
  20. 20. Slide No. 20 • 2014-04-14 Wolfgang Thielemann Document List view All documents related to Endometriosis alert:
  21. 21. Slide No. 21 • 2014-04-14 Wolfgang Thielemann Links in Document List View Link to corresponding enhanced alerting system record with: • Key content • Chemical structures Original full-text Link to corresponding patent record from Thomson World Patent Index Original PDF Patent DB
  22. 22. Slide No. 22 • 2014-04-14 Wolfgang Thielemann Document view added key content
  23. 23. Slide No. 23 • 2014-04-14 Wolfgang Thielemann browse records + chemical structures Document view + chemical structures
  24. 24. Slide No. 24 • 2014-04-14 Wolfgang Thielemann Alerting System Main Navigation: Search
  25. 25. Slide No. 25 • 2014-04-14 Wolfgang Thielemann Search result list with filter options Search for: “endometriosis”
  26. 26. Slide No. 26 • 2014-04-14 Wolfgang Thielemann Logic “OR” within on topic and “AND” between topics OR AND Text mining generated, standardized added value terms allow easy filtering: Faceted Filter Options
  27. 27. Slide No. 27 • 2014-04-14 Wolfgang Thielemann Faceted Filter Options
  28. 28. Slide No. 28 • 2014-04-14 Wolfgang Thielemann Email push service 4
  29. 29. Slide No. 29 • 2014-04-14 Wolfgang Thielemann Email push service
  30. 30. Slide No. 30 • 2014-04-14 Wolfgang Thielemann Summary 5
  31. 31. Slide No. 31 • 2014-04-14 Wolfgang Thielemann Advantages of proprietary Patent Alerting System • Grouping of patents into project specific folders (full flexibility in defining project relevant parameters like indication, target, technology etc.). Faster evaluation by project team members due to high relevance of hits • Use of already existing, proprietary terminology for search and analysis of patents • Systems maintained by patent information professionals who are experts in the field of patent specific sources, challenges, pitfalls as well as scientific text mining analysis • System open to link and exchange information to other internal platforms via automated protocols / APIs • Fast provision of representative chemical structures allow quick evaluation of novel chemical space as well as easy download & processing of real chemical structures
  32. 32. Thank you! Acknowledgements: Selected images were licensed from: 123RF©

×