Abstract
This study examines the role of online health communities in Brazil dedicated to cannabis treatments for chronic diseases as platforms for evidence-based activism. Using a mixed-methods approach, the research combines qualitative analysis with computational techniques, including Latent Dirichlet Allocation (LDA) topic modeling, to analyze six online groups from WhatsApp and Facebook. Key themes emerging from the analysis include treatment per pathology, treatment effects, access barriers, peer support, and advocacy efforts. The findings reveal how these communities act as epistemic networks, where patients and caregivers co-produce knowledge by sharing personal experiences and engaging in dialogue with healthcare professionals. This study highlights how online health communities transform experience sharing into structured evidence, enabling collective action to address barriers such as limited access to cannabis-based treatments. It underscores the potential of digital platforms to empower patients, foster collaboration with healthcare professionals, and influence health governance.

Abstract
Online health communities enable patients and caregivers to share experiences, seek advice, and collaboratively generate knowledge about treatments and condition. However, accessing relevant information often proves challenging due to platform limitations like insufficient search functionalities. A previous study identified key topics discussed in Brazilian online health groups centered on cannabis treatments for chronic diseases. Building on these findings, this study introduces a proof-of-concept chatbot designed to enhance access to the collective knowledge within these communities. The chatbot prototype, built using Google Dialogflow, was tailored to provide contextually relevant, accurate, and user-friendly responses. A user study involving 38 participants evaluated its performance, showing high user satisfaction, task completion rates, and trust in the information provided. The results highlight the chatbot’s potential enhance knowledge accessibility, promote patient engagement, and support evidence-based activism by organizing and disseminating community-generated content effectively. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Abstract
Text readability is vital for effective communication and learning, especially for those with lower information literacy. This research aims to assess Llama 3's ability to grade readability and compare its alignment with established metrics. For that purpose, we create a new dataset of article lead sections from English and Simple English Wikipedia, covering nine categories. The model is prompted to rate the readability of the texts on a grade-level scale, and an in-depth analysis of the results is conducted. While Llama 3 correlates strongly with most metrics, it may underestimate text grade levels.

Abstract
Text simplification converts complex text into simpler language, improving readability and comprehension. This study evaluates the effectiveness of open-source large language models for text simplification across various categories. We created a dataset of 66,620 lead section pairs from English and Simple English Wikipedia, spanning nine categories, and tested Llama 3 for text simplification. We assessed its output for readability, simplicity, and meaning preservation. Results show improved readability, with simplification varying by category. Texts on Time were the most shortened, while Leisurerelated texts had the greatest reduction of words/characters and syllables per sentence. Meaning preservation was most effective for the Objects and Education categories.

Abstract
Entity linking is an important task in medical natural language processing (NLP) for converting unstructured text into structured data for clinical analysis and semantic interoperability. However, in lower-resource languages, this task is challenging due to the limited availability of domain-specific resources. This paper explores a translation-based cross-lingual entity linking approach using GPT models, GPT-3.5 and GPT-4o, for zero-shot machine translation and entity linking with in-context learning. We evaluate our approach using a Portuguese-English parallel dataset of radiology abstracts. Our results show that chunk-level machine translation outperforms sentence-level translation. Moreover, our translationbased approach to cross-lingual entity linking of UMLS concepts outperformed the multilingual encoder method baseline. However, the in-context learning entity linking approach did not outperform a translation-based approach with a dictionary-based entity linking method.