Cloud Hosting

By Clinical Leader Editorial Staff

The pharmaceutical and biotech industries have experienced multiple sea changes in the last few years, as emerging science promises advanced therapeutics that require new, complex clinical trial designs. But has the technology that supports clinical research kept up with the science behind it? In 2021, Patrick Nadolny, global head of clinical data management at Sanofi, made several predictions about clinical trial technology in Designing a Data Management Strategy for the Future. Clinical Leader editorial staff recently caught up with Nadolny to revisit his predictions, examine current trends in clinical trial technology, and imagine what innovations will shape the industry in the next few years.

Nadolny predicted that data management would evolve into clinical data science due to the influx of data sources and emerging protocols with new trial designs. This evolution has already begun, and its progress hinges on four main pillars: risk-based methodologies, AI, complex protocol designs, and DCTs.

How Are Risk-Based Methodologies Changing Trial Life Cycle?

First, risk-based methodologies have transformed study conduct. The 2016 ICH-E6 revision on good clinical practice demonstrated the urgency of adopting and adapting to risk-based study monitoring across functions, which is anticipated to be re-enforced in the upcoming revision.1Going beyond study conduct, the ICH E8 revision advocates for quality by design, focusing on both critical quality factors and operational feasibility.2This requires clinical data scientists earlier involvement to design appropriate data collection and review strategies. Likewise, the EMAs recent reflection paper on the use of AI in clinical research is fully risk-based, providing early insights on risk levels and validation needs for AI solutions during the drug development lifecycle.3

Moreover, todays clinical trials generate enormous amounts of data, with a growing proportion being eSource, and Nadolny states that current technology is not as efficient in managing the 5Vs of clinical data as required (volume, velocity, variety, veracity, and value), but its progressing in the right direction.4Also, risk-based approaches interconnect with the other three pillars. For example, companies may turn to AI to manage the large amounts of data generated from these studies to identify risks or automate repetitive activities. Likewise, these protocols are complex and often employ DCTs or hybrid technology in conjunction with risk-based approaches.

In the past, companies used the same processes for multiple studies, Nadolny explained. "But now, the methods developed for one study may not translate to the next. Several years ago, choosing a trial design was like finding a recipe in a cookbook. But now, companies are given a list of ingredients and must decide the best way to combine them to maximize each element while creating a unified whole."

What Is The Role Of AI?

AI is also revolutionizing the pharmaceutical and biotech industries. Nadolny explained that although AI is not a new technology, the advent of generative AI platforms like ChatGPT has accelerated investment and interest.

AI, especially generative AI, can be useful across the entire lifespan of a trial, from recruitment to post-study data management, Nadolny stated. Previously, AI utility was limited to simpler tasks such as identifying data patterns or reading medical images, but generative AI could ultimately also create study plans, read protocols, and suggest potential root causes of a study problem. Additionally, it can assist recruitment by better identifying potential participants. With DCTs, it can review data and identify complex data anomalies from wearable technologies. AI solutions will develop rapidly in the next few years as large and small pharmaceutical and biotech companies discover ways to automate or radically transform their processes to improve study timelines.

According to Nadolny, there are not enough qualified people in the industry to manage the vast volume of information generated by todays studies, and AI is necessary to create insights from billions of data points and various data sources. Many processes could be automated to reduce workload and inefficiencies in clinical trials, expediting data management. However, AI may not be the best solution for every study currently underway. For ongoing large or long-term trials, such as many oncology studies, integrating AI after the study starts would not necessarily save time or effort due to the complexity of transitioning to a new model. However, for new studies, implementing this technology from the beginning can accelerate timelines and enable new processes that companies can use as a template for future studies.

How Will Protocols Become Patient-Driven?

Meanwhile, complex protocol designs present more challenges to clinical data scientists. Umbrella, basket, and adaptive trial designs are just a few study protocols that can accelerate drug development but create operational complexities for data management. For example, evaluating one therapy across several indications simultaneously in a basket study is more efficient than examining one indication at a time but adds complexity to data collection across multiple medical conditions. Likewise, adaptive study design improves the predictability of study outcomes by allowing sponsors to adjust dosages and timing based on individual participant responses to the IP. However, collecting and managing the interconnected web of data these studies generate is an intricate process that today's platforms aren't fully equipped for. Too often, information is siloed, and separate systems must be integrated and reconfigured for each design adaptation, adding time to the study.

Operationally complex protocol designs may also result from the desire to meet requirements from regulatory bodies, such as ensuring patient diversity and patient centricity. Nadolny emphasizes that being patient-driven is a complex issue and is not the same as being patient-centric. For example, decentralized clinical trial procedures appear patient-centric because they allow subjects to participate remotely. However, mandating telehealth technology or wearable devices may burden some participants who would rather go to a traditional clinical setting to receive care.

On the other hand, a truly patient-driven trial would be much more complex. A patient-driven trial considers these factors and creates a flexible operational design that best fits each participant's lifestyle. Subjects would choose between participating in the trial remotely, in-person, or a hybrid mix. However, this hypothetical trial design is not yet possible to deploy efficiently with today's technology because it would create protocols that are too complex to pragmatically operationalize. The push for greater patient centricity and growing recruitment needs may drive the industry toward achieving highly adaptable, customizable trials. Nadolny predicts that technology will adapt to make such trials possible in the next two to five years.

Are Fully Decentralized Trials Possible?

In addition to meeting patients needs, the DCT trend that took off during the COVID-19 pandemic shows no signs of slowing down. Nadolny expects DCTs to continue to rise in popularity in response to other types of emergencies, such as wars or natural disasters, which can otherwise halt studies. By decentralizing trials and running global studies, companies can pivot when factors beyond their control shut down sites. However, Nadolny points out that currently, no single platform can run a fully decentralized pivotal clinical trial, and DCT technology is often a patchwork of integrated solutions. He expects the industry to invest heavily in creating new systems to accommodate the unique needs of DCTs.

The pandemic forced the industry to rethink how we work to become more resilient and adaptable," Nadolny explained. Weve learned to balance the risk of implementing new technology against the risk of doing nothing. The industry is changing, if slowly. There's a divide between the old ways and the new, and we're still coping with legacy systems while investing in the future."

In his 2021 predictions, Nadolny stated that data managers weren't fully utilizing emerging technologies because decentralized workflows and shifting protocol designs were still very new, and users faced challenges adjusting to the new normal. Currently, however, data management is catching up with industry trends.

"Everything that's happened in the past few years has forced us to adapt and maximize all our solutions," Nadolny explained. "Data management has evolved significantly. However, we're still putting patches on things and learning what we can leverage regarding new protocol designs and technologies. We still have room to improve, especially regarding DCT support, but we're moving in the right direction."

What Is The Future Of Data Management?

In 2021, Nadolny stated that clinical data management needed to evolve into clinical data science. That evolution is still necessary and is ongoing. As risk-based methodologies, AI, complex protocols, and DCTs continue to shape the industry, data management platforms must adapt to meet their needs. In addition, resiliency to emergency crises has become an imperative to infuse into our daily operations. Therefore, technology must adapt to ongoing clinical research changes at the study, country, site, and even patient level. At the same time, technology should allow for greater patient-driven solutions by giving subjects more opportunities to participate on their terms. Nadolny is optimistic that these changes will benefit companies, sites, and patients.

The industry will continue to show resiliency as we walk the tightrope between adaptive, highly complex protocol designs and patient centricity, Nadolny states. Well also see the gap close between clinical research and regular standards of care so that we dont have different processes for running a study and caring for patients. The technology we need to cope with todays challenges is still emerging, but were closer today than we were three years ago.

Here is the original post:

The Future Of Clinical Data Science Is Closer Than It Appears - Clinical Leader