Until recently, most industries had not yet adapted data-driven approaches for making decisions, driving sales, and improving customer experiences. But, for a long time, the Life Sciences industry has been rooted in data, given that the compilation of safety, efficacy, and quality information all depends on data collection, curation, management, analysis, and interpretation.
As with any digital transformation initiative, changing organizational culture is required for successfully adopting a data-driven approach. Specifically, there are four major pillars to keep in mind for good data management: Strategy and Governance, Standards, Integration, and Quality.
Most importantly, in order to be data-driven, an organization must embrace data as a corporate asset. This requires creating a data strategy that accommodates for collection, harnessing, cleansing, reconciling, and managing vast amounts of data, thus making sense of data through analytics. In turn, this will improve decision-making capabilities and improve operational efficiencies. A good data strategy should ensure that all data initiatives follow a well-defined approach that is both repeatable and measurable. Uniformity and consistency are critical for ensuring that all enterprise solutions for leveraging data follow commonly understood processes across the organization.
Driving good data strategy also requires strong data governance. Based on our experience with large data transformation initiatives across multiple industries, we recommend:
- Gaining executive sponsorship to establish data strategy and data governance with appropriate authority.
- Creating a cross-functional data governance team that includes business users and analysts, data stewards, data architects, data analysts and application developers.
- Establishing strong governance structures that address data stewardship, proactive monitoring, and periodic data reviews.
- Defining data stewards for various business domains and developing processes for review and approval of data elements.
Good data management processes first require defining data standards. There are several standards within the Life Sciences industry to define data and document structures. For example, clinical data must be captured, managed, and reported in Study Data Tabulation Model (SDTM) format. Similarly, EMA is currently devising product profile data standards, such as Identification of Medicinal Product (IDMP), while eCTD is becoming a standard for submission of documents across many health authorities. As part of the ongoing efforts within the DIA RIM Working Group in developing the RIM Reference Model, we encourage using a core model for defining RIM related data attributes. All of these standards and models set the stage for a common understanding of data across sponsors, health authorities, vendors, and other stakeholders’ contribution towards the eco-system. We also recommend keeping the following in mind:
- Try to create business glossary data standards, nomenclature, and adopt controlled vocabularies established by WHO, MedDRA, ISO and others.
- Establish and/or leverage strong Master Data Management for product-related data. There should only be one authoritative source for core product definition.
- Define data validation and business rules that are embedded into the existing processes and systems.
Data standards and consistent approaches for enterprise data management simplify data integration from various functional areas, processes, and systems. One of the objectives behind the RIM Reference Model is for sponsors to follow a common model that lends itself to better integration among entities throughout mergers and acquisitions. Such common models simplify data migration from one system to another.
Also, common data structures allow for data integration across multiple functional areas. For example, if a common product definition is followed by various functional areas, such as Clinical, Regulatory, Safety and Quality, then it’s much easier to create an integrated line of sight for developing a product from R&D all the way to manufacturing. We recommend that organizations prioritize the following:
- Eliminate data duplication by integrating data wherever possible through interfaces to other systems.
- Reduce point to point integrations – instead, rely on services, like data virtualization, and advanced mechanisms to access existing data without transformation.
- Implement data unification approaches so data is reconciled and available for analytics and insights.
Data can only be a strategic asset once there are enough processes to support, govern, and manage data quality. Using these considerations, good data quality can be achieved and maintained.
- Define the right governance structures.
- Embed data quality in performance objectives of individuals with incentives and/or penalties.
- Establish data quality metrics with periodic audits.
- Improve the overall user experience by enforcing data entry rules, constraints, automatic naming conventions, and alerts/notifications for upcoming or overdue data entry tasks, etc.
These four pillars form the basis for a strong data-driven organization that can be leveraged to generate meaningful insights for better decision-making. With these pillars in mind, embarking on a data journey can be done strategically with governance, strategy, and quality leading the way.