Paper Plain: Making Medical Research Papers Approachable to Healthcare Consumers with Natural Language Processing

Abstract

When seeking information not covered in patient-friendly documents, like medical pamphlets, healthcare consumers may turn to the research literature. Reading medical papers, however, can be a challenging experience. To improve access to medical papers, we introduce a novel interactive interface-Paper Plain 1 -with four features powered by natural language processing: definitions of unfamiliar terms, in-situ plain language section summaries, a collection of key questions that guide readers to answering passages, and plain language summaries of the answering passages. We evaluate Paper Plain, finding that participants who use Paper Plain have an easier time reading and understanding research papers without a loss in paper comprehension compared to those who use a typical PDF reader. Altogether, the study results suggest that guiding readers to relevant passages and providing plain language summaries, or "gists, " alongside the original paper content can make reading medical papers easier and give readers more confidence to approach these papers. CCS Concepts: • Human-centered computing → Empirical studies in interaction design; Interactive systems and tools; User studies.

From Paper to Card: Transforming Design Implications with Generative AI

“…Not limited to these approaches of supporting general paper-reading practices, generative AI models have also been shown to be effective in augmenting domain-specific papers. For example, August et al proposed a tool called Paper Plain [6], which leverages an LLM to lower the barrier to consuming medical knowledge by augmenting medical papers with features such as section summaries and passage-specific question-answering.…”

Relatedly: Scaffolding Literature Reviews with Existing Related Work Sections

“…I find the Related work sections helpful for this, not so much for the paper list. " (P14); (3) gives the right amount of relevant context around papers: "Gave a lot more context and resulted in fewer papers being read -had to open fewer pdfs. " (P13); and (4) helps track progress and prioritize what order to read things in: "can help me to keep track of my pace of learning about the topic. "…”

“…The opportunity here is that seeing different perspectives from authors of different papers describing a research topic and each other's work has the potential of avoiding lay-people overly trusting a single piece of evidence [18]. In this case, a future version of Relatedly could help not only link unfamiliar terminology to definitions [24] or summarize paragraphs in plain language [3], but more importantly also surface agreements and disagreements between prior works and their levels of uncertainty while helping users build confidence and trust about their learning could be especially important [18].…”

“…Other tools use visualization to connect and cluster papers [38] using metrics based on citations or semantic embedding vectors, such as Connected Papers. 3 However, it can be hard for users to comprehend the underlying meaning of complex graphs and clusters as automatic clusters often conflate multiple dimensions [25]. As a result, when timely survey papers are not available, scholars still need to examine many individual papers and try to figure out the latent themes and connections between them to conduct literature reviews [57].…”

Threddy: An Interactive System for Personalized Thread-based Exploration and Organization of Scientific Literature

“…Therefore, prior work has explored ways to mitigate this cost. One thread of research focused on cross-referencing within a single document to improve readability (e.g., [3,19,49]). A sub-thread of this is helping users more efficiently navigate to and from different parts of a table and the corresponding snippets of text [4,36,37].…”