Globally, regulatory agencies are asking pharma companies to use real world data and real world evidence (RWE) in their decision making across drug discovery, development and commercialization. RWE can provide insights for decision support in regulatory, medical and commercial affairs, but there is a complex landscape of stakeholders involved in regulatory policies and guidelines that relate to the use of real world data, such as digital health data, including government agencies, NGOs, and health tech assessment agencies.

One pharma company wanted to identify the stakeholders across the RWD landscape to help understand digital health data policies and pinpoint the key opinion leaders and influencers in this space. The global pharma company wanted to perform an assessment of the stakeholders influencing the intersection of digital heath data policy (DHDP) and real-world evidence guidelines pertaining to drug development. The company was interested in understanding government agencies, non-government organizations, conferences, search engines, and trade groups involved in influencing and shaping real world and digital health policies.

Identifying these stakeholders would help the company understand the issues driving decision-making around DHDP, and the role their own internal experts could play in shaping these conversations.

However, conducting such an analysis manually would take months of research performed by a team of experts, and they still would not likely uncover all the most relevant documents to comprehensively outline the complexity of the ecosystem.

They decided to work with IQVIA to apply innovative technologies in order to discover the relevant KOLs. IQVIA experts in regulatory science and strategy and real world data worked with IQVIA technical experts to use a combination of AI technologies (natural language processing and knowledge graph) to create an effective robust solution that could provide answers to a range of relevant questions in a matter of minutes, rather than days or weeks.

Natural Language Processing (NLP) is an AI technology that can extract key information from unstructured text, rapidly and effectively.  Knowledge graphs are conceptual models of various entities and their relationships within a network that enables users to quickly find and connect heterogenous data by means of standardized ontologies. The use of NLP enables key data from unstructured text to be extracted, standardized, and input into the graph, as well as relevant structured data. Once data is stored as a knowledge graph, this can be used to visualize the data as a network which can be more intuitive than the conventional modelling of data in rows and columns. Moreover, structuring data in a knowledge graph enables data scientist to conduct analyses that would otherwise be extremely time-consuming or impossible.

Developing the regulatory policy graph

Search criteria were set by the regulatory and RWD experts, in collaboration with the pharma researchers. Once search criteria were set, the NLP team built and trained AI algorithms on what to search for, where to look, and how to dig deeper based on citations and references within the text. The NLP team used their platform to draw structured meaning from the data based on context, frequency, and connections between stakeholders. The team standardized the output using defined ontologies, then passed the results to the Knowledge Graph team who ingested the NLP output into a graph model.  Having the data modelled in a knowledge graph enabled visualization of the entities and their relationships to graphically depict the key-opinion-leader landscape as well as providing a basis to apply graph analytical algorithms to score the importance of documents and stakeholders based on their links to other documents and stakeholders.

After they were deployed, the combination of NLP data and graph technology enabled the team to identify patterns and illuminate complex relationships in minutes. The graph visualization is color-coded to represent different policies, stakeholders, domains, and what connects them, providing decision makers with a graphic representation of relevant stakeholders and documents. Interactive features allow users to click through different categories to understand the influence and activity of stakeholders and documents in each subgroup. This ease-of-use was vital for decision makers, who had limited experience with data mapping.

Further, the algorithms also tracked citations and references to locate stakeholders who may be quietly influential but not identified in the initial landscaping assessment. The citation tracking helped the pharma company see which of their competitors were being cited and had influence in the digital health data policy space. By using artificial intelligence (NLP and knowledge graph technologies), IQVIA was able to run this analysis in less than 20 minutes after initial landscaping assessments were completed.

Understanding digital health data policies for strategic insights

The insights provided the company with understanding of DHDP in real-world evidence regulation creation and positioned the company to drive strategy discussions. The company used the insights to hone a list of the most influential stakeholders to engage, and in some cases to identify which associations were less influential than expected.

These insights are now informing the company’s outreach and helping company leaders identify opportunities to participate in conversations and planning for how real-world evidence will evolve with future policies and regulations.