Pharmaceutical research is a dynamic field that is always changing as new technologies transform traditional methods. The use of artificial intelligence (AI) to assist in the creation of lay summaries (LS) of clinical trial outcomes is one such breakthrough. In order to ensure that patients, their caregivers, and the general public may easily access and understand study results, LS are crucial in creating transparency. The rapid advancement of AI technology presents both new challenges and substantial benefits that must be carefully investigated while creating LS. Lay summaries attempt to make clinical research findings more comprehensible for audiences that are not scientists by simplifying complex medical material. AI can increase LS drafting work by decreasing manual labor (i.e., time and resources). But people can make mistakes and misunderstand things. For example, using data from websites like ClinicalTrials.gov may not offer the complete and accurate LS (Edwards, 2025).
The Rising need for Brief Summaries
Because clinical trials provide patients with more treatment options and aid in scientific research, they are a crucial component of evidence-based medicine. Even though they are very important, there still are not enough accessible resources available to teach patients and the public about different trial options.
This is the problem that has been known for a long time. Trial registries such as ClinicalTrials.gov are currently the main source of information for physicians and people looking to investigate their possibilities for a trial. Despite this, these platforms frequently employ extremely technical jargon that is difficult for the general public to understand, even though they are publicly available. Access to clinical trial data is extremely beneficial for cancer patients, but they are also more likely to have misinterpretations. The amount of information and severity of their illness can be tiring. Even though things like multimedia interventions and plain-language summaries have tried to make things clearer, an adaptable and customized solution has not yet been achieved. In light of this, the increasing use of generative artificial intelligence (AI) in healthcare offers a chance to close this communication gap (Goldstein & Krukowski, 2023).
The implications for APAC leaders and sponsors with U.S. bases are both morally and strategically significant. A brand's repute is improved by direct communication. Patient-centered communication increases interaction and certainty. Short summaries assist in preventing misrepresentation in the digital age, where misconceptions can spread quickly. Because it is easy to get to, a wide range of readers will be able to understand the science that affects them.
Why Interpreting Clinical Trial Data Is challenging
Interpreting clinical trial data can be difficult, as it requires detailed research designs, deep statistical analysis, and specialized technical reporting. Trials commonly include subset analyses, multiple study groups, and defined objectives that require careful evaluation. Without training it is hard to understand statistical terms like "risk estimates" and "p-values." Differences between trial participants and actual patients and incomplete reporting and publication bias are additional factors that may influence the reporting of results. Clinical trial reports with massive data volumes are much tougher. Accurately interpreting clinical trial results is difficult for researchers, doctors, and the general public due to these issues combined.
How Artificial Intelligence Converts Complexity to Clarity
A common subclass of generative AI text models, large language models (LLMs) have demonstrated great potential in a variety of medical applications, ranging from summarizing medical literature to carrying out tasks that support doctors in clinical decision-making. An innovative and engaging approach to enhancing patient comprehension and involvement is the incorporation of AI-powered text production into clinical trial instructional materials. The study looks at how LLMs, in particular GPT-4, can improve cancer clinical trial education materials in order to assess LLM potential in this area. Specifically, the study created patient-friendly summaries related to different oncologic clinical trials using informed consent forms (ICFs) from ClinicalTrials.gov. The authors used two methods to create ICF summaries in order to achieve this aim. Asking the LLM to produce a brief summary of an ICF in a single step was the first technique, known as direct summarizing. The second approach, sequential summarizing, used a series of steps to methodically improve the summary's accuracy and clarity. The authors present the new discovery that patient-friendly summaries that were considerably simpler to comprehend than the original ICFs were generated by both direct and sequential summarization techniques. One hopeful development in patient participation is the capacity of LLMs to translate intricate medical terminology into easily understood summaries. Widespread usage, however, requires addressing issues like content authenticity, AI hallucinations, and authorities auditing. The authors note that institutional review boards typically conduct additional human evaluation of even manually created patient-facing recruitment and consent documents in light of this (Waters, 2025).
Using LLMs for Patient-Centered Design and Communication
Although artificial intelligence-based systems have the potential to enhance trial design and implementation, patient participation is still essential to trial success. Thanks to legislation changes like the US's CURES Act, patients can now access their medical records. But because of its complications, this data can be troubling. Large language models, or LLMs, have the capability to turn this setback into an opportunity. It is convenient to create more readable, organized representations from unstructured clinical data. For example, NYUTron, which was created to understand complicated EHR records, can assist in matching patients with the most suitable trials. By adjusting content to the language and community preferences of potential participants, LLMs have also shown early promise in improving ways to communicate. This goes beyond basic translation to accomplish cultural adaptation. Breaking down detailed consent forms can help potential volunteers in understanding the risks, advantages, and study procedures. In clinical trials this kind of personalization is very important because it helps individuals to understand what their roles are and what will happen as a result of their participation. But when using LLMs in areas where patients are present, like informed consent, you should be careful, as there will be misinformation generated by AI. For these technologies to integrate smoothly into clinical protocols, we need regulatory oversight, and thorough testing is essential (Badani et al., 2025).
Figure 1: Programmed for Automated Machine Learning Clinical testing (Badani et al., 2025)
Transparent Clinical Trial Communication: Its Significance in the US and APAC
In order to promote confidence and transparency, knowledgeable and dubious patients in the US want precise explanations of clinical trial outcomes, including risks and limits. AI speeds up the delivery of understandable lay summaries by sponsors, enhancing public access and enhancing business reputation. Growing clinical research activity in the Asia-Pacific area needs communication that is adapted to a variety of linguistic and cultural norms. AI makes it possible to localize tone and complexity effectively while preserving regional scientific coherence.
Participation of research sponsors
The goals, endpoints, design, and interpretation of the study are the responsibility of the research sponsors. Their participation is essential for proper interpretation of complicated data and for making sure that Lay Summaries (LS) appropriately reflect trial outcomes. The isolated development of LS by outside parties carries the danger of being misunderstood and losing crucial context as trial results become more widely available to the public. It has been shown that the lack of sponsor monitoring can result in misinterpretation or the deletion of crucial information in the LS, even while enhanced accessibility mechanisms can encourage equity in the transmission of information (Edwards, 2025).
Cultural sensitivity and subconscious bias
Unusual results may arise from intentional or unintentional biases in training information. When artificial intelligence algorithms are trained on large datasets that may not fully reflect the diversity of cultural backgrounds, the generated content may lack cultural understanding and comprehension. AI has the capacity to replicate and even magnify these biases, producing distorted summaries that undermine the fairness of data disseminated to patients and the general public (Edwards, 2025).
Suggestions for utilizing AI effectively in the creation of lay summaries
Stakeholders with the following extra AI expertise and experience may be crucial at different phases, even though standard operating procedures and resources at organizations may differ:
● Experts in AI development and training
● Experts in health literacy
● Compliance and legal teams
● Medical writers and LS
● Standardized glossaries and templates
● integrity of the data inputs
● Prompt engineering
● advanced AI architectures
The skill of the humans participating in AI's training, prompting, supervision, creation, and modifications of lay summaries (LS) determines how well AI generates LS. To ensure adherence to best standards and maintain quality and accountability, all reviewers should continue to apply their core duties, abilities, and credentials in the LS process even with the incorporation of AI tools.
Figure 2. Proposed modifications to the process flow in accordance with the guidelines of Good Lay Summary Practice (GLSP) (Edwards, 2025)
Conclusion
There are pros and cons of using AI-generated lay summaries, which means that planning and execution must be done very carefully. AI can make things better and reach more people, but humans need to make sure that data provided by AI is correct and follow the rules. Successful execution will require constant observation, evaluation, and development. Ultimately, integrating AI into LS development needs a balance between innovation and control to make sure that each summary meets the top level of quality, accuracy, and transparency, increasing patient and public confidence and clarity.
References
Badani, A., De Moraes, F. Y., Vollmuth, P., Chung, C., & Mansouri, A. (2025). AI and innovation in clinical trials. Npj Digital Medicine, 8(1), 683. https://doi.org/10.1038/s41746-025-02048-5
Edwards, K. (2025). Considerations for the use of artificial intelligence in the creation of lay summaries of clinical trial results. Volume 34.
Goldstein, C. M., & Krukowski, R. A. (2023). The importance of lay summaries for improving science communication. Annals of Behavioral Medicine, 57(7), 509–510. https://doi.org/10.1093/abm/kaad027
Waters, M. (2025). AI meets informed consent: A new era for clinical trial communication. JNCI Cancer Spectrum, 9(2), pkaf028. https://doi.org/10.1093/jncics/pkaf028
EN 
cn 
de 
jp 
