Clinical Trials Do Not Have a Data Problem, They Have a Decision-Readiness Problem

Our 2 companies have worked side by side on clinical studies for more than 7 years. One of us builds technology where trial data begins.

The other works where that data becomes analysis, evidence, and ultimately the basis for clinical and regulatory decisions. Watching studies cross that boundary, again and again, has convinced us of one thing: The industry’s hardest data problem has moved.

Clinical research has made remarkable progress in collecting data. Studies today draw on electronic data capture, electronic clinical outcome assessments, wearables, central laboratories, imaging, decentralized sources, and other data streams that barely existed a decade ago. Capture is faster, richer, and more flexible than ever.

Increasingly, analytics are becoming faster too. Artificial intelligence (AI), automation, and advanced analytical approaches are making it possible to identify signals and interrogate clinical data while a study is still in motion.

That is real progress. But real-time analytics and real-time decision readiness are not the same thing.

A signal may be generated instantly. The harder question is whether the underlying data are sufficiently clean, reconciled, standardized, traceable, and understood for a development team to act on it with confidence.

For many studies, this is becoming the real constraint: not how quickly data can be captured or an analysis generated, but how far the data must travel across systems, transformations, teams, and handoffs before someone can confidently act on it. We think of this as the distance between data capture and trusted decision.

The view from the point of capture

The advances in data capture are genuine, and they have changed what a trial can see. But every advance has also exposed the seams downstream.

Each new data source adds an integration point, and each integration point is a place where quality, timing, and meaning can drift apart if no one is watching the whole picture. The lesson we have drawn on the technology side is one of humility.

Faster collection does not automatically produce faster decisions. Without the right downstream design, it can simply produce a larger backlog, sooner.

The view from the point of decision

From the analysis and evidence side, this gap often looks less like an execution problem and more like a design problem.

Sponsors invest heavily in capturing data and then too often treat everything that follows — cleaning, reconciliation, standards, mapping, review, statistical analysis, and submission datasets — as downstream execution to be handed off and optimized primarily for cost.

In our view, that framing misses much of the value.

These activities are not simply the cost of having collected data. They are the process through which data becomes evidence, and that process can be designed well or badly long before the first patient is enrolled.

The most effective teams we see architect this path backward from the decision.

What will the safety committee need to know? What will the program lead need? What evidence will eventually need to withstand regulatory scrutiny? And when will each answer be required?

Once those questions are defined, the data architecture, standards, review processes, and analytics can be designed around them.

On studies in which the flow from capture through mapping, review, and analysis is specified from the start, we have seen review meetings answer questions from data the team already trusts rather than trigger another cycle of queries and reruns.

In those situations, database lock becomes less of a discovery event and more a confirmation of what the team already knows.

Regulation is opening the door

The regulatory environment is also moving in this direction. Modern good clinical practice, most visibly the International Council for Harmonisation’s updated E6(R3) guideline, increasingly emphasizes quality by design, proportionate risk management, fit-for-purpose systems and processes that support reliable decision-making throughout the data lifecycle.

That matters because speed and compliance should not be viewed as opposing forces. The opportunity is to design clinical data processes in which speed, quality, traceability, and reliability reinforce each other.

Regulation is not preventing continuous decision readiness. Increasingly, it is defining the conditions under which it can be trusted. A real-time analytical signal generated from data that is still being reconciled, standardized, or interpreted is not yet decision-ready evidence.

The next step is therefore not simply making analytics continuous. It is making more of the underlying data-to-evidence lifecycle continuous as well.

What shared studies have taught us

First, treat the path from capture to decision as one connected system with clear end-to-end accountability, rather than a relay race between technology platforms, data teams, statisticians, and service providers that meet primarily at handoffs.

Second, design backward from the decision. Ask what needs to be known, by whom and when, and architect the data path accordingly.

Third, measure what matters. Database lock is an important but lagging indicator. More revealing measures include how quickly discrepancies surface and are resolved, how early analysis and submission data sets take shape, and how often a decision is delayed by data that technically already exists.

And fourth, do not mistake automation for readiness.

AI can accelerate many parts of the journey from capture to decision, but it does not eliminate the need for structured, traceable, and well-understood data. If anything, as automation accelerates, that foundation becomes more important.

Where the next gain comes from

The industry is already moving toward real-time data and real-time analytics.

The next frontier is more demanding: Can we make trusted evidence available at the speed at which decisions need to be made?

That will not come from technology, services, standards, or AI independently. It will come from connecting them — designing capture, integration, standardization, review, analysis, and evidence generation as one continuous system rather than assembling them through a series of handoffs.

We hold that view with some confidence because we have spent 7 years observing and testing it together on real studies, from opposite ends of the data path.

When the path is designed as a whole, data problems surface earlier, review becomes more continuous, decisions spend less time waiting for data, and database lock increasingly confirms what the development team already knows.

The goal is no longer simply faster data, it is trusted data that are ready for the decision when the decision needs to be made. That, we believe, is the next evolution of clinical development.

About the Authors

Amir Malka, CEO, Bioforum, and Philip McCrea, CEO, Viedoc.