Generative AI Holds the Key to Transforming Trial Design

Opinion
Article
Applied Clinical TrialsApplied Clinical Trials-12-01-2023
Volume 32
Issue 12

While still in its infancy, generative AI will continue to be integrated into clinical operations.

Tobias Guennel, PhD, SVP Product Innovation/Chief Architect, QuartzBio, part of Precision for Medicine

Tobias Guennel, PhD, SVP Product Innovation/Chief Architect, QuartzBio, part of Precision for Medicine

In 2023, artificial intelligence (AI) has been ubiquitous, so much so that “AI” was named the most notable word of the year by the dictionary publisher Collins. While use of AI has been accelerating at an incredible pace in the past couple of years, it has been used for decades in the biopharmaceutical industry, primarily for modeling, to identify drug targets, or point to potentially new indications or patient populations for existing drugs. However, these applications rely heavily on predictive AI, which harnesses historical data to make predictions and recommendations for the future.

But what is really propelling the industry forward is generative AI, which uses models or algorithms to create various types of content using learned patterns based on the data on which it was trained.

Generative AI holds extraordinary promise for transforming clinical research, from how trials are designed to how the data are analyzed. One of its core applications in clinical trial design is data management. Generative AI enables the collection of vast amounts of data from the public domain, such as public data repositories and scientific literature, and synthesizes it in a concise way to inform clinical trial designs. Our focus on innovation within our AI-enabled biomarker intelligence solutions is to reimagine interaction patterns of key stakeholders to significantly reduce the time needed from posing a scientific question to generating the right insight via easily consumable information. This ensures that the right people have access to the right information faster than ever before.

For example, in the oncology space, researchers typically use publicly available data across a multitude of clinical indications coupled with proprietary data generated through the course of clinical research. These data can help identify patient characteristics and biomarker profiles that could, in turn, inform novel trial designs. The ability to combine these assets of public and proprietary data to model future clinical trials is a core principle of AI.

However, before modeling and analytics can be used to simulate a trial on a particular drug target within a particular patient population, it needs to be consolidated into a high-quality dataset that can be used to inform simulations. For example, this can be achieved by generating synthetic data that model different patient subsets. To do so, it is important to manage the data and transform it into a standardized data model. AI can be used in various ways to create that high-quality data foundation, and generative AI can be utilized to make that data foundation more accessible. Our clients use our AI-enabled solutions to integrate these data in ways that are useful for researchers and subsequently enable researchers to converse with their data to drive key insights.

So what is slowing its adoption and use in clinical trial design? For one, the data being generated are highly sensitive, such as patient information and intellectual property, making data privacy and security paramount. The need to guarantee data privacy leads to more conservative approaches on how to adopt new technologies. As with any new technology, best software development lifecycle and data security practices as well as a robust model training and validation approaches need to employ. In clinical trials, researchers also need to make sure that the data and models being used have a strong foundation to avoid the “garbage in, garbage out” problem. If the data at the starting point are poor quality, the data generated will also be poor, potentially leading to faulty outputs. As a result, there must be greater scrutiny in making sure that the process to build algorithms and models are based on strong AI approaches.

Although still in its infancy, generative AI will continue to make its way into day-to-day clinical trial operations. In the coming years, we will see a much larger, and faster, adoption of this new technology. Transforming how end users interact with software or technology platforms is imperative and will continue to evolve in clinical trial design. The possibilities for adapting and utilizing new technologies in the clinical trial space are limitless—we have yet to see what can be done. 

Tobias Guennel, PhD, SVP, product innovation/chief architect, QuartzBio, part of Precision for Medicine

Related Content
© 2024 MJH Life Sciences

All rights reserved.