Decentralizing pharma data

Sept. 14, 2023
How data mesh can help the pharma industry free data from silos

The pharma industry can unlock a wealth of advantages by tapping into the capabilities of data analytics.

Pfizer, for example, uses data to create an efficient manufacturing and supply network that streamlines the way it delivers medicines to patients. Roche employs data analytics to advance its personalized medicine initiatives, provide holistic care and drive better patient outcomes. But there’s a catch.

Given that many global pharma companies have teams in different regions, establishing a centralized data system to power analytics throughout their organizations becomes challenging. This is primarily due to the diverse regulatory environments in each region, as well as the distinct suborganizational goals, priorities and IT budgets. Global teams own and manage different data hubs, leading to data silos.

At its core, a data silo occurs when there is a lack of connectivity between data sources, meaning that other groups in the same organization struggle to easily or fully access data. When critical data becomes trapped, users are forced to rely on partial or incomplete information, negatively impacting the decision-making process across the entire organization. According to a recent survey, 68% of knowledge workers — pharmacists, engineers, architects, scientists and academics — report that not having visibility into cross-functional projects adversely affects their work.

Data silos exist throughout the modern pharma industry, from databases to shared network folders. They even exist between research colleagues at the same company. And even though data silos often emerge within large pharma organizations, any pharma business — regardless of size — may find itself facing data silos if there is no carefully-planned data management strategy.

Now is the time to take a closer look at the impact of data silos on the pharma industry and how data mesh, a modern approach to data management based on decentralized data architecture, can help companies address this issue.

The problem with data silos

Effective cross-functional collaboration plays an important role in pharma organizations, as each department brings a unique set of skills, knowledge and perspectives to the table. That’s why collaboration is crucial when it comes to improving information flow, identifying potential risks before they escalate, reducing bottlenecks and adapting to ever-changing market demands and regulatory requirements.

However, a recent Aspen Technology survey highlighted that nearly half of the senior decision-makers from drug development or manufacturing companies agree that data silos significantly hinder cross-functional collaboration and efficiency within their organizations. For this reason, removing data silos and barriers to data integration and sharing can help pharma companies enable faster processes, improve overall efficiency and save valuable time and resources.

In addition to that, pharma companies accumulate massive volumes of data throughout the drug development life cycle, from preclinical research, clinical trials, manufacturing and marketing. On average, phase 3 clinical trials alone produce approximately 3.6 million data points. This is three times more than the data gathered during late- stage trials a decade ago, according to a 2021 study from Tufts CSDD. The same report also suggested that clinical trial designs are poised to become more complex and generate even greater data volume and diversity in the future.

The heart of the matter is that many data silos are not consistent with other data sets due to infor- mation stored in various systems or scattered throughout different departments. When data from one silo does not perfectly align or match with data from another silo, companies cannot integrate and analyze information effectively, leading to inaccurate results or incomplete insights.

On top of that, regulatory agencies, such as the U.S. FDA, have strict requirements for the quality, accuracy and integrity of data submitted in drug applications. Therefore, inconsistent data makes it difficult for pharma organizations to comply with regulatory require- ments, which increases the chances of rejection and results in fines and even legal consequences.

For example, in 2012, GSK pleaded guilty and agreed to pay $3 billion to settle criminal and civil charges, covering issues such as failure to report safety data, unlawful drug promotion and alleged false price reporting practices.

Eliminating data silos and enabling seamless information flow can also allow pharma companies to significantly improve data accuracy and ensure that sen- sitive data is well-protected. This, in turn, helps reduce the risk of unintentional data breaches, increases compliance efforts, streamlines regulatory inspections and prevents potential noncompliance penalties.

“Even the unintentional release of sensitive medical information is a serious breach of consumers’ trust,” highlighted J. Howard Beales, director of the Federal Trade Commission’s Bureau of Consumer Protection back in 2002 after Eli Lilly had settled charges concerning a security breach. “Companies that obtain sensitive information in exchange for a promise to keep it confidential must take appropriate steps to ensure the security of that information.”

How data mesh can help pharma

The solution to problematic data silos lies within modern decentralized data architectures like a data mesh. Acting as a unifying force, a data mesh is a trans- formative approach to data management, creating a data-first community that seamlessly connects data providers with data consumers in a secure manner.

A data mesh’s value goes beyond just fostering efficient communication: It also helps reduce complexity by granting users access to relevant information instead of all available datasets, which is particularly beneficial for pharma companies navigating large amounts of data. Thanks to this capability, teams are able to make quick decisions and enhance overall operational efficiency.

However, implementing a data mesh approach may be easier said than done, as there are numerous technical challenges to overcome. For starters, following the domain-driven, self-service data mesh infrastructure can involve a significant investment in hardware and software resources, requiring databases, messaging systems and application programming interfaces.

Creating a mindset change within the business domain where the members are the owners of the data and responsible for its quality and processing capability presents another hurdle. This is primarily because traditional business hierarchies often place data-related responsibilities in the hands of IT depart- ments or data specialists. But regular communication about the significance

of data ownership and its direct correlation to better decision-making, drug development and patient outcomes can help pharma companies emphasize the importance of this transformation and speed up the process.

The fruits of implementing a data mesh can be significant, especially in the pharma domain, where data is more diverse, global and covered by more regulations. Creating ownership fosters bottom-up accountability and a data-first mindset that can be crucial to unlock exponential business value from data.

Build a data mesh strategy

Here are eight steps pharma companies can follow in order to establish and implement a data mesh strategy.

Step 1: As-is assessment

Organizations must gain a deep understanding of their existing business challenges, data infrastructure, processes and systems. Businesses can achieve this by defining key objectives, performing a business analysis, assessing data infrastructure, and cataloging and categorizing all collected data. The as-is assessment needs to be aligned according to the various business domains. This way, they can identify and prioritize the data gaps that require immediate attention.

Step 2: Identify business domains

Each domain — whether research and development, clinical trials, manufacturing, supply chain, sales or marketing — has its own data needs and stakeholders. Therefore,

pharma organizations need to make sure to identify the different business domains existing within their company, focusing on the ones highlighted in Step 1 as the most important areas to address.

Step 3: Formulate cross-functional domain teams

This involves working with domain experts, data engineers, data scientists and analysts. These teams will be responsible for managing the data within their respective domains and aligning it with an organization’s overall strategy.

Step 4: Assign data product owners

Data product owners on each domain team can focus on the quality, availability and usability of the data products within their domains.

Moreover, they can collaborate with other domain teams to define data product boundaries and ensure consistency and interoperability. These owners would also own compliance of their data with the specific regula- tory frameworks governing it.

Step 5: Implement decentralized data infrastructures

There are several decentralized data infrastructures that support the data mesh approach, such as data vaults. This typically involves implementing a data warehouse that serves as a centralized storage repository for the data products.

Step 6: Enable data product discovery

Develop mechanisms for data product discovery and access. This may include a data catalog or data marketplace that provides information about available data products, their owners and relevant metadata, which is important to ensure data privacy, security and compliance. Variyng regulatory and organizational priorities can result in different departments having their own data hubs, which leads to data silos. Not being able to access information seamlessly impacts collaboration within a pharma company, compromises the accuracy of data and even affects compliance efforts.

As a data mesh approach securely connects data providers and users while facilitating streamlined communication, it enables pharma companies to overcome these hurdles. While challenges still exist, the benefits of adopting a data mesh approach can be substantia 

Step 7: Foster a data-centric culture

In order to promote a data-driven culture, companies can start by encouraging collaboration among employees. By doing so, team members can learn from each other’s experiences, best practices and challenges, accelerating the learning curve. Companies can provide training programs and resources to assist employees in developing data literacy skills.

Step 8: Monitor and iterate

Regularly monitoring the data mesh implementation and seeking feedback from domain teams and data users are essential practices. Based on the valuable insights received, organizations can continue to iterate and improve data products, infrastructure and processes. Given that pharma companies operate in an environment influenced by regulatory changes, clinical advancements, and shifting patient expectations, regular evaluations can enable them to maintain agility and stay relevant.

The bottom line

By harnessing the power of data analytics, pharma companies can open up numerous opportunities, such as accelerating drug development, improving patient safety and streamlining business operations. 

Varying regulatory demands and organizational priorities can result in different departments having their own data hubs, which leads to data silos. Not being able to access information seamlessly impacts collaboration within a pharma company, compromises the accuracy of data and even affects compliance efforts.

As a data mesh approach securely connects data providers and users while facilitating streamlined communication, it enables pharma companies to overcome these hurdles. While challenges still exist, the benefits of adopting a data mesh approach can be substantial.

About the Author

Sourabh Banthia | Lead Consultant, Advisory Team, Gramener

Contact Sourabh on LinkedIn