New digital therapeutic approach to health literacy can enhance patients’ engagement with educational materials

About 93% of U.S. adults access the Internet, and 80% of those search for health information, despite low levels of proficiency in health literacy. In a new study, researchers developed an automated approach to assess how easily understandable patient educational videos are for individuals looking for information on diabetes. The study’s findings offer insights for content creators and health care organizations on improving users’ engagement with online materials by developing a digital therapeutic approach that uses evidence-based digital health technologies to change behavior persuasively.

Conducted by researchers at Carnegie Mellon University (CMU), Arizona State University (ASU), and Michigan State University (MSU), the study appears in the Journal of Medical Internet Research (JMIR).

“With the vast amount of health information available in multimedia formats on social media platforms such as YouTube and Facebook, billions of people across the globe are accessing health care information via social media without any way to verify the accuracy, understandability, or relevance of the content,” says Rema Padman, Trustees professor of management science and healthcare informatics at Carnegie Mellon’s Heinz College and principal investigator of the NIH funded project, who coauthored the study. “In this context, there is an urgent need and a unique opportunity to design a way to curate online health information using multiple criteria to meet the health literacy needs of a diverse population.”

Providing access to high-quality health information and educational materials for patients is essential for empowering patients, improving health and cost outcomes, and building societal resilience. But limited health proficiency is a challenge worldwide; in the United States, only 12% of adults are considered proficient in their ability to interpret health information meaningfully.

In this study, researchers developed a human-in-the-loop, augmented intelligence approach that focuses on the interaction between humans and algorithms on YouTube—the largest video-sharing social media platform. The approach combines guidelines from the Patient Education Materials Assessment Tool (PEMAT) with features extracted from online videos for patients.

It also uses annotations of the videos by domain experts and co-training methods from machine learning to assess the understandability of videos on diabetes and classify them. With this information, researchers then examined the impact of understandability on several dimensions of viewers’ engagement with the videos.

The study collected nearly 10,000 YouTube videos on diabetes—among the most prevalent chronic condition in many parts of the world—using search keywords extracted from a patient-oriented forum and reviewed by a medical expert.

The analysis demonstrated that the co-training classification model, which combined machine learning with expert input, performed strongly. In addition, higher levels of understandability had a positive effect on viewers’ engagement, yielding more views, likes, and comments, and boosted the likelihood of expert recommendations for patient education. These results point to the importance of improving video understandability for enhancing patients’ engagement with educational materials on contextually relevant health-related topics, potentially advancing the health literacy of individuals and populations.

The study’s approach may be more widely applicable across various health domains, say the authors. The methods and principles could be adapted to other chronic, acute, and infectious health conditions, such as cardiovascular disease and cancer, and to broader patient education contexts, such as medication adherence and patient safety.

Among the study’s limitations, the authors note that the PEMAT is not designed for user-generated content but for materials produced by health care organizations, so the PEMAT criteria may require adaptation or extension to YouTube videos in evaluating sub-criteria (e.g., whether the materials used for illustration are uncluttered, whether the technical quality of the video is satisfactory). In addition, the study relied heavily on four physicians’ evaluations of patient education materials, which poses risks of evaluator bias.

“Currently, digital technologies for public health literacy and patient education are limited, lack scalability, and do not fully use the vast amount of publicly available health information found online and on social media platforms,” explains Xiao Liu, assistant professor of information systems at ASU’s W.P. Carey School of Business and co-PI, who coauthored the study. “Providing a strong open platform provides a credible alternative to the vested interests of private organizations with proprietary technologies, which will lead to innovations in new data collection devices, digital platforms, and technologies in the context of health literacy initiatives.”

Adds Anjana Susarla, Omura-Saxena professor of responsible AI in the Department of Accounting and Information Systems at Michigan State University’s Broad College of Business and co-PI, who coauthored the study: “Our findings can offer a path toward patient education and empowerment, as well as improved health literacy of the population, by providing clinicians and patients the ability to easily retrieve understandable and relevant video-based information on health education.”

The study was funded by the National Library of Medicine of the U.S. National Institutes of Health.