he process of designing clinical studies today entails the manual compilation of the protocol document. Several people work on building the protocol by passing document versions back and forth, and spending hours sifting through past protocols, scientific literature, and regulatory guidelines to inform the new study design. This approach comes with several limitations:

• Long timelines
• Inconsistencies and lack of standardization
• Increased risk of errors
• Lack of guardrails and controls that could prevent future protocol amendments
• Maintenance of multiple versions of the document
• Constrained collaboration and communication
• Challenges to ensure regulatory compliance
• Manual downstream processes based on the protocol document.

It can take nearly a year to design a new protocol!

In a blog post last year, I listed protocol design as one of the drivers of avoidable costs in clinical trials. And a recent article in the DIA Global Forum pointed to potential savings on the order of millions of dollars if protocol design were to be digitized and enhanced by artificial intelligence (AI). In the opening paragraph, Kimberly Tableman wrote, “Swapping digital for paper has proven to be the springboard for operational transformation across dozens of industries from financial services to manufacturing. Yet protocol design—the building block for clinical research—remains rooted in paper, belaboring an already long, slow process.” Inspired by the argument for digitizing protocol design made by Kimberly, I offer a follow-up.

The value of digitalization of the protocol design process should not be underestimated. This short video from TransCelerate summarizes the challenges of the document-driven approach to protocol design and the promise that comes with its digitization. That’s why I embraced the opportunity to get involved with TransCelerate’s Digital Data Flow (DDF) initiative earlier this year. I’ve been excited to participate in the DDF’s Solution Collaboration Forum as part of a broader effort that aims at modernizing clinical trials by enabling a digital workflow, starting with establishing a universal study definitions model (USDM).

Consider the transformative potential of a common data model that enables seamless communication across downstream systems supporting clinical trials end to end. Envision its power when built into a digital study builder, so that the study protocol is interoperable by design. What if an electronic case report form (eCRF) could be automatically generated based on information from the digital protocol; if visit schedules could be created with a proverbial click of a button by passing data from the digital protocol to the scheduling systems; or if the trial design domains in the study data tabulation model (SDTM) could be autopopulated with data from the protocol?

The advantages of a USDM-enabled digital study protocol builder are not limited to those realized downstream from the study design itself. Protocol digitalization, especially when augmented with AI, would make the protocol design easier, more efficient, and streamlined because it would include:

• A digital tool where collaborators across functions—medical, toxicology, regulatory, statistics, ethics, and others—can work in a sequence of orchestrated handoffs or asynchronously to design a study protocol,
• Guided and gated workflows with business rules that take protocol creators through the design process step by step, and help prevent errors and omissions,
• A smart tool that can populate protocol fields based on built-in logic and information entered in earlier steps of the workflow, that enforces standardized data formats, and enables value selection for standardized data fields,
• Automated version control with an auditable change log and tracking,
• A virtual, generative AI-powered assistant that can analyze past protocols, relevant scientific literature, and regulatory guidelines to suggest the narrative for selected sections of the protocol; and that makes recommendations to the study design to prevent avoidable protocol amendments in the future,
• A tool with an approval workflow that outputs a compliant document and supports an electronic regulatory submission.

Readers may find it informative to review the evolving list of potential solutions that utilize the USDM and that have been published (without endorsement) by TransCelerate in the DDF Solution Directory.

Digitalization of the clinical study protocol is within reach, and organizations are already experimenting with the application of generative AI for protocol authoring and generation. Take for example one biotechnology company’s successful efforts to implement GPT across enterprise, including to inform dosing decisions in study design, or a study at Georgia Tech that demonstrated feasibility and efficiency gains from leveraging GPT4 in protocol authoring. Finally, collaboration efforts such as the Solution Collaboration Forum under the auspices of TransCelerate’s DDF initiative are an incredible opportunity to exchange experiences, discuss challenges, and share solutions; to evangelize USDM as an enabler of interoperability across clinical trial systems; and to train and prepare technology vendors for universal adoption of a common data model we desperately need to make clinical trials more efficient, effective, cost-effective, and faster.

Combining cutting-edge technologies such as generative AI and modern digital platforms with a widespread adoption of the USDM by sponsors will drive significant efficiencies in clinical trial design and execution.