SCOPE 2024: Driving Efficiencies in Trial Strategy, Oversight With Generative AI

Image credit: AndSus | stock.adobe.com.

Advances in automation with large language models (LLMs) can advance the development of clinical trial strategy from days to minutes, according to an expert panel at the Summit for Clinical Ops (SCOPE) Summit 2024 in Orlando, Florida. Panelists Wendy Morahan, senior director, Clinical Data Analytics, IQVIA Technologies, and Wing Lon Ng, director, AI Engineering, IQVIA, also elaborated on how trial oversight dashboards are generated with natural language.

Ng kicked off the session with a discussion on ways to improve clinical research, explaining that users can harness technology to help review documents and data at a much faster rate.

“When needed, we can ask an engineer to identify and replace sensitive information. This offers new ways of streamlining data management,” Ng said in his opening remarks. “Data extraction is something that many people have already used with ChatGPT because it is able to distill large amounts of information faster. You can find the key extracts that you want and can help get the insight you want much easier. We can use chatbots to help us answer questions related to a specific domain and allow users to understand content much quicker.”

After Ng finished providing examples, Morahan began by discussing the use of conversational artificial intelligence (AI) in study oversight.

“I think that study oversight—specifically sponsored oversight of the study—is extremely important to all of us in clinical trials and industry interests,” Morahan said. “At the end of the day, they're responsible for patient safety. For me, one of the key use cases is having extremely robust and flexible capabilities for sponsor oversight.”

Morahan also noted that AI shows a significant amount of potential in areas such as automation, detection, and predictive analytics.

“This is where we’re able to take a massive amount of historical data that you have at hand and leverage it to understand the current event that's happening in our trial, and predict what that might mean about the future,” she explained.

Additionally, Morahan discussed several examples of generative AI being used for study oversight. This included code generated programming in areas such as data quality checks and conversational analytics.

“What I mean by that is simple. You enter a query, and it’s just a fancy way of saying question. Then in real time, getting a response back from the system. Finally, we have a text paragraph describing an answer and that might give visualizations. This is basically conversational analytics or conversational AI,” Morahan continued.

Morahan then provided conference attendees with a demonstration in the use of conversational AI. As part of the demonstration, she asked questions such as when a patient’s earliest and latest visit date occurred.

“You do start learning that you have to be specific,” she said. “How you ask the question is really important.”

Concluding the presentation, Morahan offered some final thoughts on the use of AI in this space.

“We’re all talking about it, but it’s not a silver bullet. It’s not the answer to every challenge that we have in the world, in other industries, or in life sciences. But it is extremely important. And it’s something that is said to be very informative,” she said. “I don’t think we can talk about this topic without saying, ‘Hey, everybody, don't forget, we have to think about ethical, responsible, safe use of AI and generative AI.’ But I think that we're going to do them and then we're going to disrupt how we do things. I'm looking forward to see what happens over the next couple of years.”