Speed to IDMP Readiness A case study in data mining

Transcription

1 Speed to IDMP Readiness A case study in data mining sales@cunesoft.com 2016 Cunesoft GmbH December 8, 2016 Corporate Overview Cunesoft Jim Nichols

2 Transparency is not a choice! #regulatoryglasshouse 2016 Cunesoft GmbH Corporate Overview Cunesoft

3 Data Requirements Now Drive Regulatory Compliance Paper Binders Before 2000 PDF PIM & SPL 2010 xevmpd IDMP Cunesoft GmbH Corporate Overview Cunesoft

4 Why wait to be penalized? i.e. EMA will charge up to 5% of annual gross revenue for IDMP non-compliance 2016 Cunesoft GmbH Corporate Overview Cunesoft

5 Manual Data Mining - Time intensive - Not scalable - Costly Automated Data Mining #regulatoryglasshouse - Results within minutes - Multiple use cases - Immediate compliance 2016 Cunesoft GmbH Corporate Overview Cunesoft

6 Corporate Facts Global Life Sciences Customer Headquarters: Munich, Germany Regional Offices: Paris, Bangalore, Princeton Industry Focus: Regulatory Software for Life Sciences Venture Partners: HTGF, Bayern Kapital, Occident Group Rainer Schwarz Founder and CEO Daniel Koppers Founder and CTO cune-distiller Extract - Enrich - Evolve Regulatory Operations Platform Designed for Life Sciences Preconfigured IDMP Library SmPC s, XEVMPD s, ectd data mining Sophisticated yet simple to use Build on industry best practices Private SaaS or On-Premise 6

7 Agenda The Foundation of the Challenge The Hurdles A Data Mining Overview Case Study Findings Beyond IDMP 2016 Cunesoft GmbH Corporate Overview Cunesoft

8 The foundation of the challenge 8

9 SPOR Substance, Products, Organizations and Reference Master Data 9

10 IDMP Pathway to Compliance Today March UNDERSTAND ISO IG s June September 17 ASSESS First draft of regional EMA IG s September December December EVALUATE Consultation period for draft EMA IG s IMPLEMENT Release of EMA IG s January June 18 PREPARE 6 months implementation timeline July June SUBMIT IDMP is mandatory and you have 12 months time to submit your IDMP submissions After June 19 COMPLY You would be non-compliant if not having submitted IDMP 10

11 IDMP Iteration 1 79 Fields (Expected) Iteration 1 scope overview per data category: Medicinal Product: Marketing Authorization: Organisation: Marketing Information: Pharmaceutical Product: Ingredients: Package Information: Indication: Total: 28 Fields 10 Fields 05 Fields 05 Fields 06 Fields 09 Fields 12 Fields 04 Fields Fields 11

12 IDMP Iteration 1 Data Source Analysis Data Mapping Expected Mandatory Fields 79 Medicinal Product 28 = 36 % Marketing Authorization 10 = 13 % Organization 5 = 6 % Marketing Information 5 = 6 % Indication 4 = 5% Pharmaceutical Product 6 = 8 % Ingredient 9 = 11 % Package Description 12 = 15 % Data Sources 1. SmPC 2. xevprm 3. SPL 4. ectd 3.2.P 5. Substance Data Base 6. Product Data Base 7. RIM System 8. Document Management System MedDRA G-SRS EV-Codes SNOWMED ATC EDQM Personal databases... 12

13 IDMP is far more than a regulatory task Product master data management, Product registration ID management, Data collection, etc. 1. Regulatory Affairs (Headquarter and Affiliates) 2. CMC Substance data management, Substance registration tracking, Substance information collection, etc. Manufacturing 3. Manufacturing license data, Manufacturing information, physical product characteristics collection 6. Quality Control 5. Supply Chain Medical 4. Clinical Particular management, including coding for Indications, Undesired Effects, Contra Indications, etc. 13

14 The hurdles still remain! 2016 Cunesoft GmbH Corporate Overview Cunesoft

15 What challenges are ahead? IDMP Iteration 1 field definition While the EMA has published a list of the likely IDMP Iteration 1 fields, there are still a number of discussion around several attributes between industry and regulators including medicinal product name splitting and addition of manufacturing/establishment information. Data preparation scope uncertainty IDMP Iteration 1 is one clear scope of data preparation and curation. But while extracting data from 3.2.P within ectd s related to your product would it make sense to curate additional IDMP attributes at the same time? Do you perceive real value in collecting the information internally as well? Identifying the regulatory single source of truth When starting data collection for IDMP, the SmPC seems like one big data source, but detailed analysis shows that in some cases substance data might be better extracted from a specific database or 3.2.P.1 while the nature and contents of the container is still better described in the SmPC than within 3.2.P.7. The data collection has to be conducted on several data sources with a system that allows to concatenate and curate from several sources. Final implementation guidelines missing While IDMP Iteration 1 data field requirements are well known by now, the business rules on mandatory and optional values as well as the value structure are not. This poses a risk during data preparation that needs to be managed. There are things that will be low risk (e.g. extension of XEVMPD concepts like name splitting), but others at higher risk (e.g. Risk of Shortage of Supply where there is no pan-european definition). Substance data quality & registration process While the FDA and Health Canada already provide an operational test instance of G-SRS (ISO implementation), the EMA does not provide a system for substance data quality and integration testing yet. While this would make the identification of data gaps possible, the other missing key component is a description how the substance registration process will work and who will be the registrar into G-SRS as well as clear identification of roles and responsibilities. Referentials (including OMS) DB not ready yet While the current implementations of OMS and RMS are being finalized with user acceptance testing being conducted at the moment, the lack of complete data sets does hinder data collection, data curating and integration efforts by the industry and vendors. 15

16 IDMP Area IDMP Attributes cune-distiller Source Document Medicinal Product Combined Pharmaceitical Dose Form SMPC, xevmpd, SPL Medicinal Product Additional monitoring indicator SMPC, xevmpd Medicinal Product Orphan Designation Status SMPC, xevmpd Medicinal Product Name (Med.Product) SMPC, xevmpd, SPL Medicinal Product Invented Name Part xevmpd Medicinal Product Scientific Name Part xevmpd Medicinal Product Strength Name Part SMPC, xevmpd, SPL Medicinal Product Pharmaceutical Dose Form Part SMPC, xevmpd, SPL Medicinal Product Formulation Part SMPC, xevmpd, SPL Medicinal Product Intended Use Part SMPC, xevmpd, SPL Medicinal Product Target Population Part SMPC, xevmpd, SPL Medicinal Product Container or Pack Part SMPC, xevmpd, SPL Medicinal Product Device Name Part SMPC, xevmpd, SPL Medicinal Product Trademark or Company Name Part SMPC, xevmpd, SPL Medicinal Product Time/Period Part SMPC, xevmpd, SPL Medicinal Product Flavour Part SMPC, xevmpd, SPL Medicinal Product Classification System SMPC, xevmpd, SPL Medicinal Product Classification System Value SMPC, xevmpd, SPL Marketing Authorisation Marketing Authorisation Number SMPC, xevmpd Marketing Authorisation Country xevmpd Marketing Authorisation Authorisation Status xevmpd Marketing Authorisation Authorisation Status Date SMPC, xevmpd Marketing Authorisation Procedure Identifier/Number SMPC, xevmpd Marketing Authorisation Procedure Type (e.g. MRP/DCP) SMPC, xevmpd Marketing Authorisation Country (national authorisation) xevmpd Marketing Authorisation Marketing Authorization Number (national authorisation) xevmpd Organisation Location Address SMPC, xevmpd, SPL Indication Indication Text SmPC Indication Indication as "Disease/ Symptom/ Procedure" SMPC, xevmpd, SPL Indication Co-Morbidity SmPC 40 of 79 IDMP Iteration 1 fields are mostly located within unstructured data sources such as SMPC, XEVMPD, SPL or ectd. cune-distiller extracts these within seconds and does code mapping for MedDRA, G-SRS, EV- Code etc. Indication Intended Effect SmPC Pharmaceutical Products Administrable Dose Form SmPC Pharmaceutical Products Unit of Presentation SmPC Pharmaceutical Products Route of Administration xevmpd Ingredient Ingredient Role SMPC, xevmpd, SPL, 3.2.P.1 Ingredient Substance SMPC, xevmpd, SPL, 3.2.P.1, 3.2.P.3.2 Ingredient Strength Range (Concentration) SmPC, xevmpd, SPL, 3.2.P.1, 3.2.P.3.2 Package description Package Description SMPC, SPL, 3.2.P.7 Package description Package Item (Container) Type SMPC, SPL, 3.2.P.1 Package description Package Item (Container) Quantity SMPC, SPL, 3.2.P.1 Package 2016 description Cunesoft Material GmbH Corporate SMPC Overview, SPL, Cunesoft 3.2.P.1, 3.2.P.7 16

17 End to End IDMP Data Management cune-idmp Gateway Connector cune-distiller cune-idmp MIGRATE: Get data out of documents and into data systems. STAGE: Merge all relevant IDMP data. Apply IDMP business rules. Prepare submission output. Export to MDM or RIM. 17

18 Configurable Framework Functional Description MUST HAVE SOLUTION Data Mining Unstructured data outside of systems (SmPC s, xevprm s, etc.) Structured Data inside of systems (RIM/ CMC data, clinical studies, etc.) Data Mapping Controlled vocabulary mapping (EV Codes) External database mapping (MedDRA, GSRS, ATC, WHO, etc.) Internal database mapping Data Consolidation Business rules driven data consolidation Define leading system Data Cleaning Workflow driven quality assurance process cune-distiller Data Maintenance SmPC data maintenance via structured authoring Compare various data sources with same attributes Data Export Create manual output in XLS, XML, CSV, JSON format Push / Pull integration into target system Library Management Subscribe predefined public libraries Create personal / private libraries (NLP, Neuronal, Rules Based) 18

19 Data Mining Can Deliver R Results 19

20 A Typical SmPC... Only 17 pages long! How long would it take you to find the information? 20

21 Extracting the Information Into Correct Elements How long would it take you to enter the information? And then encode it? 21

22 Automating Extraction Keep the information logically organized 22

23 Intelligently Finding the Information Understand the origin of the extraction 23

24 Automating Encoding Encode to relevant systems 24

25 Automating Encoding - MEDRA 25

26 Automating Encoding G-SRS 26

27 Enhancing QA & Corrections Easily make changes 27

28 Streamlining the Entire Process 28

29 From Possibilities to Reality 2016 Cunesoft GmbH Corporate Overview Cunesoft

30 Data Mining Case Study Results 1200 SmPCS Processing time per SmPC: 30 seconds QA Time per SmPC: minutes Manual* Automated Conversion 1.5 Hours.008 Hours 1200 SmPCs 1800 Hours 10 Hours 225 Days 1.6 Days QA of Results 300 Hours 300 Hours 37.5 Days 37.5 Days vs 39.1 *Manual data extraction assumption: 60 minutes data mining & 30 minutes data mapping (EV-Codes, G-SRS, MedDRA) 30

31 Can You Really Afford to Wait? + Manual 1 FTE 9 FTE 24 FTE Automated Time*: 225 days 25 day 7 days 1 day Cost*: 100% 100% 100% 30-50% QA Time*: 37 days 37 day 37 days 37 day Cost*: 100% 100% 100% 100% Quality: 100% 100% 100% 100% VS Saving Time Total: 262 days 62 days 44 days 38 days 85% reduction in time 31

32 More Than Meets the Eye For Now! 32

33 More Use Case Examples - BEYOND IDMP Automated etmf Metadata Tagging A seamless etmf process that extracts data from clinical trial documents (i.e. physician CV s etc.) Automated Document Redaction EMA Policy 70 requires the redaction of confidential information within clinical study reports, protocols etc. Automated creation of document abstracts Tag relevant information within documents and automatically create abstracts Automated DMS Metadata Tagging Data mine a DMS for metadata and reuse for automating other processes (i.e. ectd creation) Substance analysis versus G-SRS Identify if all relevant substances are captured within G-SRS & identify if preferred substance terms are being used internally. Data Base Quality Assurance Identifying data inconsistencies across multiple data stores and clean data automatically. Data & Document Consistency Synchronize data changes back into documents via structured authoring methods. Reduce Manual Efforts by 90% Ensure Compliance 33

34 You have proven that we have entered a new area of digitizing the regulatory world Astra Zeneca, Representative of Information Technology Department 2016 Cunesoft GmbH Corporate Overview Cunesoft

35 End to End IDMP Management cune-distiller IDMP Data Staging Data Curation Services cune-idmp Will YOU be ready? Europe: +49 (0) USA: