Democratizing data will transform biopharma R&D. It starts with the people.

We are familiar with a core problem facing biopharma: it typically takes about 10 years and more than $2 billion to bring a new drug to market. Only 10 percent of molecules tested in clinical trials reach the market. What's more, this timeline and cost haven’t really budged for decades.

This problem is complex and needs to be addressed from multiple angles:

Target validation, lead identification, and lead optimization: We need to fail faster and increase the probability of success.
Clinical trials: We need to get better at finding target populations, running creative designs, and simplifying logistics––and thus move more quickly.
Clinicogenomics: We need access to new kinds of data so we can better understand patients and disease.

…and more.

Key to enabling each of these changes is the democratization of data. That is, we need to make the vast quantities of data that biopharma companies have broadly and easily available to researchers.

But that’s not just a matter of wrangling data or building new tools; it will require researchers to fundamentally change the way they work to be data-first. It will require organizations to embrace being digital (not just digitizing). And it will involve every part of an organization: people and culture, platforms, and (of course) data.

In this piece, I’ll speak to these key components of the transformation to a data-first approach. But first, let’s take a step back and look at the big picture.

What can data-first R&D look like in biopharma?

Let’s start with a very visible example: in the winter of 2020, scientists mapped the genome of a novel coronavirus. Within two days, researchers had used that map to design vaccines for the virus, and within 10 months, two of those vaccines had undergone clinical trials, received emergency authorization, and were being provided to healthcare workers and high-risk individuals.

Admittedly, the COVID vaccines also benefited from regulatory tailwinds, federal funding, and unusual urgency, none of which is the norm in day-to-day R&D. But even more crucial was the data-first approach behind the scenes. This ranged from using computational immunology models to rapidly identify targets on the virus to optimizing selection of clinical trial sites with Real World Data (RWD) to identify current virus hotspots to numerous other links in the chain of vaccine development.

This approach is markedly different from what happens today in most biopharma R&D settings. While research scientists are among the most data-driven people in the world, they tend to view data first through the lens of evidence generation rather than as the primary tool to fast-forward research by rapidly identifying and prioritizing hypotheses before going into a “wet lab.” This excerpt from Nature illustrates the difference:

“…Peter Howley, of Harvard University, … asked why certain strains of human papillomavirus are associated with cervical cancer and others are not. Bioinformatics data showed that equivalent proteins in different human papillomavirus strains bind to a different constellation of host proteins in the cell. Howley hypothesized that these differences could explain why some strains cause cancer. It allowed Howley and other immunologists to hone their hypotheses more finely before diving into expensive and time-consuming ‘wet lab work.’”

This success, like the success of COVID vaccines, was built on a foundation. The scientists were fluent in using data-first techniques (and in the latter case had even helped develop some of them). The computational models had been developed and evolved over a number of years. Pipelines existed that were already collecting and harmonizing massive datasets.

Digital-native biotechs (like the relative newcomers BioNTech and Moderna) tend to be ahead of traditional pharmas in building this foundation. But traditional pharma has an advantage or two as well, especially in the volumes of data they already have.

Let’s dive into what the transformation to a data-first biopharma R&D setup looks like. First, the foundation necessary for transformation.

Figure 1: Transformation is about people and is supported by platforms and data

Creating trust in the solutions People: the heart of transformation

The heart of this transformation, as with all digital transformations, is the people who make up an organization. We are asking people to change how they work––to change what has made them and the organization successful.

Leadership may believe in the change, but individual researchers are still accountable for delivering ongoing results. They have deep experience and expertise in how to do this. They may be curious about a data-first approach, but can’t yet tangibly envision what this change looks like for their work or how it will help them.

They may also be concerned about the investment of time, risk of failure, and perceived loss of control in working in a new way.

How do we bridge this gap and create the belief that enables people to adopt change?

Much has been written on leadership and change management in transformations. I’d like to focus on two other aspects of this particular transformation:

How do we find the right problems to solve, ones that will create belief in the new way of doing things?
How do we create trust in the insights and solutions derived from this new way of doing things?

Finding the right problems to solve

Put differently, this is a matter of knowing where to start. In a massive biopharma organization with millions of patient-years’ worth of data, where do you begin to tap that data to create new value?

When we engage in this work, we look for problems that meet the following criteria:

They provide direct, meaningful value to the users and the business.
They have strong champions who can drive vision, adoption, and investment.
They make for good success stories and demonstrations of a data-first approach.
They can be solved reasonably quickly via incremental advances.

We sometimes refer to the process of finding those problems as art of the possible. The art is in the mix of creative and pragmatic thinking. Finding the best ideas isn’t a straightforward process. The possible reflects that the best ideas will come from an interplay of different perspectives: researchers’, data experts’, platform experts’, etc.

It is important to not view this as a brainstorm. Brainstorms lead to groupthink and often mediocre solutions. Better approaches encourage independent ideas, challenges to ideas, learning from each other, and group alignment. We use mindsets and techniques from working differently and design thinking to provide this structure and make the process effective.

Note that this approach doesn’t just help identify the right problems to solve. It also lays the foundation for adoption of the solution by creating broad buy-in and champions for the change.

One role in this we’ve found particularly important can be called the catalyst. Catalysts are individuals who can own both (1) finding the right problems and (2) championing the solutions with users. They are typically researchers themselves and so understand researchers’ needs and enjoy this group’s trust. They also have or grow understanding of the data and technology, which allows them to uniquely bridge all of the perspectives and drive the adoption of data-first culture.

Creating trust in the solutions

Researchers won’t use a data-powered tool to fuel their research if they don’t trust the data itself. There are typically three hurdles to building that trust:

Organizational silos
Researcher comfort sharing data
Researcher trust in the data and insights

Let’s look at all three.

1. Organizational silos

Large biopharmas are highly siloed into functions that often lack a culture of sharing. Different functions work independently, sourcing their own data, building their own applications and analytics, and creating their own processes. To democratize data, leaders must find a way to address this mindset. And then they have to overcome the more practical (but still daunting) hurdles of governance and technology separations.

These two shifts are among the biggest challenges in shifting to data-first work. Both require leadership from the top.

There is a silver lining, though: within these silos lies one of the biggest advantages of large biopharmas. They have enormous amounts of internal data and analytics with great potential once democratized.

Figure 2: Silos separating people, platforms & data

2. Comfort sharing data

In addition to a lack of sharing between functions, researchers can be reluctant to share their data with each other.

Most often, this reflects a concern that the data could, without the appropriate context, be misused. That’s a valid concern: there can be a lot of nuance in research data, and misinterpretation is a real risk.

There's no easy answer to this. Here are two things we’ve seen work:

Find ways to share the context along with the data.
Enable new users of the data to easily connect with the scientists who created and understand it.

Those connections offer additional benefits: conversation and collaboration among researchers can create new context that can be shared with future data users. Sharing data, in other words, isn’t just about the data.

3. Trust in insights and data

Researchers are evidence driven. They come to trust things by digging into the answer and seeing how it was created: Why should I trust this? Show the work.

A black-box answer is insufficient. So we must find ways to show the work, such as…

Making answers traceable to the raw data when possible. Show the sources of the answers. Make it simple for a researcher to pull the data and check for themselves.
Incorporating explanations of methodology in whitepapers. Ensuring that methodology is created by authorities and making it easy to contact those authorities.
Providing validation of methodologies––for example, comparisons against historic data.
Sharing code and algorithms. Allow researchers to dig under the covers. This has the side benefit of creating community and encouraging reuse and improvement like you see in GitHub.

For a data-first future, culture is key

It’s easy to think of tapping into an organization’s data as a transactional or even mechanical undertaking: switch to a better fuel, enjoy better performance.

But a biopharma organization is not a car. Biopharmas are complex entities, often made up of thousands of individual people with individual thoughts, feelings, fears, and hopes––all of which affect the transition to a data-first structure. To ensure that those powerful forces work in alignment with such a transition, biopharmas must keep culture front and center.

The very idea of democracy, of course, is that people hold the power; key to democratizing data and transitioning to a data-first way of operating is ensuring that the people in question both want this power and know how to wield it effectively.

Admittedly, this is a bigger topic than what can fit into a single article. If you’d like to continue the conversation, don’t hesitate to reach out.

Published by Rolf Russell

Knowledge not data advances drug discovery

Let’s start a conversation

Let's shape your insights into experience-led data products together.