When your research affects cancer patients, reliable AI is critical

Dr Engie El Sawaf 새 탭/창에서 열기 knows the difference research can make to people’s lives. As a Pharmacology Lecturer Assistant at Future University in Egypt 새 탭/창에서 열기, she researches neuropathy, a common side effect of chemotherapy that can be so severe that patients choose to stop cancer treatment.

“It’s extremely hard for people,” she said. “It can have a huge impact on quality of life, so if you can find solutions to decrease the pain, you can make a big difference to them during their treatment. It gives me motivation — the idea that I can do something to help these people.”

The ever-evolving nature of Engie’s field means she often needs to bring herself up to speed quickly in new areas of research. She’s found that generative AI tools, such as Scopus AI 새 탭/창에서 열기, can help bring her to the right information faster:

I’ve been working with Scopus AI for a couple of months, and it’s been great. I’d previously used ChatGPT, but it’s not a tool for scientific literature review. When I heard there was something new based on Scopus — which is the database I trust most when I’m doing a literature review — I started using it a lot.

Scopus AI draws from metadata and abstracts of Scopus documents published since 2013. The tool uses advanced prompt engineering to minimize the risk of false AI-generated information and ensuring that responses are based on current, trusted knowledge.

For a researcher, reliable information is essential. That Scopus AI is based on a curated, peer-reviewed dataset sets it apart from tools that draw on a more general dataset, Engie said:

There’s a lot of poor information that gets generated by other, less specialized tools. Even though they’re written in a way that seems true, the results are often very inaccurate or outright false. Using Scopus AI, you get information that is generated from abstracts that come from peer-reviewed journals. So mostly you get trusted information.

Engie uses Scopus AI to search the literature and generate summaries that provide an understanding of new areas.

“I’ll do a research query on Scopus to get a list of research articles and review articles and so on,” she explained. “At the same time, I will run a query on Scopus AI to get a summary. That way, before I go through all the review articles and everything on the Scopus database, I can examine the summary and understand what the key points to follow up on are.”

Researchers shaping the future of Scopus AI

Launched at the beginning of 2024, Scopus AI is at the start of its journey. The research community has contributed ideas and feedback on Scopus AI since the design stage, and Engie is one of thousands of researchers worldwide who participated in testing the product. That user community is continuing to shape Scopus AI’s future. For future iterations, Engie reflected on the possibility of a tool that provides direct answers to researcher enquiries:

“For example, if I wanted to know how many studies on a specific point had used a rat model, that would be useful — a result that wasn’t a summary but a single answer to a specific question,” she said. “Or knowing how many articles have been published since 2020 on chemotherapy-induced neuropathy and which is the highest (cited). Those kinds if developments would also be useful.” Engie also noted that it would be useful to be able to iterate on a response in the same way that ChatGPT allows, so researchers could drill down into specific elements of a summary or expand on areas of interest: “I know that we’re getting into quite advanced features here, and it’s early days, but it would be great to have these features from other AI tools combined with the Scopus dataset.”

Building a reliable generative AI tool for research community

“I am always grateful when hearing from our user community about their experience with Scopus AI, said Yoshiko Kakita 새 탭/창에서 열기, Senior Director of Scopus Product Management at Elsevier. In this case, Engie’s experience struck a personal chord:

Yoshiko noted that it was especially valuable to hear Engie’s recognition of the trustworthiness of the Scopus content, as well as her suggestions for further improvement.

“Our approach is deeply rooted in valuing user feedback,” she said. “The Scopus AI you see today has been significantly shaped by the feedback we've received from thousands of users across the globe. Some of the feedback from Engie is already on our roadmap.

“We believe that generative AI is here to stay,” she continued, noting that the team is keenly aware of the concerns that accompany generative AI, such as hallucinations, bias and privacy issues:

The research community is actively discussing how best to utilize generative AI in advancing research while addressing these concerns. We hope to contribute to advancing this important discussion by providing a reliable generative AI tool, Scopus AI, which was our major motivation when we first started building it.

Yoshiko explained that the team is committed to using AI responsibly:

We are striving to ensure that our innovations not only push the boundaries of what's possible but also maintain the utmost ethical standards and integrity. Together with the research community, we are excited to explore the vast potential of generative AI with Scopus AI.