#United States #New York #AI tools #NYU Langone #Patient Education

Enhancing Patient Education through AI Tools

In an era where accessibility is paramount, a groundbreaking study led by researchers at NYU Langone Health has revealed how artificial intelligence (AI) can drastically improve the readability of online patient education materials (PEMs). These materials are crucial for helping patients navigate their health care decisions, but many are often too complex and exceed the recommended reading level of grade 6.

The Need for Readability in Health Materials

According to the study, patients frequently find themselves overwhelmed by the sophistication of medical texts, which were originally crafted by health experts. The traditional creation of patient education materials typically caters to professional norms rather than considering the diverse literacy levels of patients. This disengagement from the patient's perspective can hinder effective communication and negatively impact health outcomes. With an average readability score of 10.7 for materials from the American Heart Association, 10 for the American Cancer Society, and 9.6 for the American Stroke Association, it’s clear that many materials are not accessible enough.

How AI Tools Optimized Readability

The study investigated the capabilities of three significant language models, namely ChatGPT, Gemini, and Claude, to rewrite these health materials in simpler language. The researchers prompted these AI tools to transform the original texts without sacrificing their accuracy. Remarkably, the adjustments led to substantial improvements — ChatGPT reduced the readability score to a grade level of 7.6, Gemini achieved a level of 6.6, and Claude reached an impressive 5.6. Additionally, word counts were significantly minimized, contributing to concise information dissemination.

Dr. Jonah Feldman, the study’s senior author, emphasized the transformative potential of large language models in health communication. “Our findings show that the integration of AI tools can enhance even professionally composed health materials, ultimately empowering patients through better understanding,” he stated.

Real-world Applications and Randomized Trials

The implications of this study extend beyond theoretical applications; the NYU Langone team is actively incorporating these AI innovations in practical scenarios. Already underway is a randomized controlled trial that utilizes AI-generated, patient-friendly summaries for hospital discharge instructions. This initiative aims to compare the effectiveness of these AI-optimized materials against conventional instructions in enhancing patient comprehension and satisfaction.

Dr. Paul Testa, a co-author of the study, noted the vast potential of AI in health care systems globally, stating, “The adaptability of AI technologies showcases how communication within healthcare can be transformed, ultimately benefiting patient experiences around the world.”

Moreover, real-world effectiveness is being validated through ongoing clinical trials, as the research team aims to produce conclusive evidence on AI tools' usefulness in improving patient care interactions. Dr. Jonah Zaretsky reiterated the importance of generating empirical data, asserting, “Verifying that AI-generated content is not only accurate but beneficial is critical for enhancing patient and family experiences.”

Conclusion

Ultimately, the research out of NYU Langone not only serves as a pivotal advancement in patient education but also illustrates the dynamic role that technology can play in healthcare. The ability of AI to create clearer, more comprehensible health information is a significant stride toward inclusive healthcare communication. By prioritizing understandable materials, healthcare systems can enhance patient empowerment and actively contribute to better health outcomes, translating complex medical jargon into support and accessible guidance for patients.