Medidata Blog
The Future of Data Linkage in Clinical Trial Evidence Generation
Arnaub Chatterjee, SVP of Medidata AI discusses the benefits, the regulatory perspective, and the future of big data and AI in clinical trials.
In the third installment of this three-part blog series, Arnaub Chatterjee, SVP of Medidata AI, joined MIT Technology Review’s Business Lab podcast to answer more key questions about the future of AI and data linkage in clinical trial evidence generation.
How will scientists, healthcare providers, and ultimately patients, benefit from clinical trial data linkage? If Medidata has the world’s largest data set, what does that mean? What can you find? What can you see?
Historical clinical trial data on such a large scale has the power to create major change in life sciences. Our data might challenge preexisting assumptions that scientists have. It might even challenge a current belief that is accepted in the industry. So, not only can historical data serve as a benchmark for a pharma company’s current performance, but it can also guide a company’s strategy as they move forward. This can have many benefits across the innovation stakeholder spectrum—from helping to reduce uncertainty, accelerating timelines, or getting to a go/no-go decision faster. AI technology can even show a sponsor if their evidence is substantial enough to run a costly Phase II or III study before the trial even begins.
If a sponsor were to build longitudinal cohorts as single patients with linked data, they could follow these patients long after the trial. This significant period of extended observation could help them answer many hard questions about their treatment, like:
- How do we evaluate long-term efficacy?
- What are the long-term outcomes for patients?
- How can we prospectively design a clinical trial that improves drug performance in the future?
This long-term monitoring is crucial in certain therapeutic areas, like CAR T, where regulatory bodies are asking for 10 to 15 years of evidence after the trial. Data linkage provides a systematic way of collecting that data for individual patients.
What is the regulatory perspective on this new data that is being generated and used for evidence submission? How are pharmaceutical companies adapting to that changing landscape?
What we have to understand is that the regulatory perspective is going to change as use cases become validated. The recent FDA draft guidance that came out in September 2021 is a critical step forward for RWD use in submissions.1 This was one of the first times that the regulatory bodies have said “If you have routinely collected data outside of the trial, this is what rigorous evidence looks like.” As technology companies and drug developers, we need to understand how to design trials effectively based on these guardrails.
As a pharmaceutical company, you have to approach the regulatory bodies well in advance. If you’re planning to propose data linkage or a synthetic control arm, you have to pre-specify this strategy in your statistical analysis plan well in advance so that regulators can vet this approach. When engaging regulatory bodies, it is always a great opportunity to hear their feedback and gain a better understanding of what kind of data you need to address the problem that you’re trying to solve.
For example, we’ve seen in synthetic control arms, which have been used for a number of years now, that there are disease areas where it makes sense. But there have also been some disease areas where the FDA has said that a traditional randomized trial would work better. I think this approach is very reasonable. Overall, it seems regulators are amenable to new approaches—we just have to implement them judiciously.
In the next three to five years, what are the opportunities and challenges ahead in drug development?
Right now, we’re probably still in the early days of looking at data and AI and saying, “Are these algorithms good or not?” We want to make sure that we create generalizable, representative, bias-free algorithms that reflect the diversity of populations. But in parallel, we should also be looking to understand the rigor required for these algorithms to augment physician or provider decision-making. Within the next three to five years, I think there is going to be a real change in the industry as we get more of this validation and qualification for some of these algorithms.
I’m also really excited about the changes happening in the regulatory environment. We’ve seen a high number of drug approvals and growing acceptance of novel methodologies, such as synthetic control arms. Although this change may be slow, I’m hopeful that, as companies like Medidata try to create these expeditious ways of getting drugs into the hands of patients, that these methodologies will be well received by regulatory agencies.
Listen to the full discussion here.