The aim of the Social Media Mining for Health Applications and Health Real-World Data (#SMM4H-HeaRD) shared tasks is to fos- ter the development and evaluation of natural language processing, machine learning, and artificial intelligence methods for analyzing health-related text from social media and other real-world data sources. For the 11th iteration, held online and co-located with ACL 2026, the workshop continued the expanded #SMM4H- HeaRD platform initiated in 2025, broaden-ing its scope beyond social media to include additional health real-world data sources such as clinical narratives and biomedical literature. The 8 shared tasks covered diverse data sources, health domains (e.g., adverse drug events, insomnia, influenza vaccine effectiveness, cancer staging, substance use), and task formulations (e.g., classification, named entity recognition, span extraction, and text generation). In total, 110 teams registered, representing 31 countries. In this paper, we present an overview of the datasets, participant systems, and performance results, providing insights into current methods for mining social media and health real-world data for biomedical and clinical applications.

Patient-generated symptom expressions are linguistically diverse, often deviating from standardized medical terminology. This paper introduces the Japanese Patient Phrase Bank (JPPB), the first automatically constructed phrase-level normalization resource for Japanese patient language. JPPB introduces an embedding-based soft labeling framework that transforms traditional one-to-one dictionary mappings into graded and ambiguity-aware associations. This framework represents a shift from word-level to phrase-level normalization in Japanese. The resource covers 7,035 phrase–term pairs across 412 symptoms. Evaluation on the KEEPHA and MedNLP-SC datasets shows that soft labels consistently improve Top-1 accuracy and better approximate gold label distributions compared with hard labels. While LLM-based normalization achieved the highest scores, JPPB provides a lightweight and transparent alternative suitable for local deployment. This work demonstrates that large-scale, automatically generated phrase banks can achieve competitive performance relative to manually curated resources and serve as practical, scalable resources for medical natural language processing in Japanese.

Recent advances in large language models (LLMs) have accelerated the NLP applications in the medical and clinical domains. However, evaluations remain limited for non-English languages, such as Japanese, where clinical corpora are particularly scarce. To address this gap, we present J-ClinicalBench, a publicly available benchmark designed to reflect realistic Japanese clinical tasks. We first created 227 expert-authored clinical documents and newly constructed five datasets for core clinical tasks. Building on these datasets, J-ClinicalBench comprises nine clinical tasks spanning clinical language reasoning, generation, and understanding. We establish baseline performance on J-ClinicalBench by evaluating state-of-the-art proprietary and Japanese open-source LLMs, providing the first assessment of their utility in practical clinical scenarios. By releasing this benchmark, we aim to foster the development and evaluation of clinically applicable LLMs in Japanese healthcare, bridging the current gap between clinical NLP research and clinical practice.

Medical concept normalization in clinical text is a fundamental technology for the secondary use of clinical data. However, constructing annotated resources for this task is challenging because annotation is both expertise-intensive and methodologically complex. As a result, a standard evaluation dataset for Japanese has yet to be established. In this study, we introduce a Japanese dataset for medical concept normalization, MedNormJ, which will be publicly available. The dataset consists of 397 pairs of medical expressions and their corresponding normalized disease names, manually curated from 96 medical documents, including case reports and radiology reports. Furthermore, we conduct comparative experiments using existing normalization approaches to benchmark their performance on this dataset in terms of both accuracy and computational efficiency. Through these experiments, we clarify the present performance level and identify remaining challenges specific to Japanese medical concept normalization.

In recent years, patient narratives have been used to understand subjective experiences that are not recorded in clinical notes. However, narratives tend to be long and unstructured, requiring summarization. However, text-based summaries often require a lot of clarification from patients and make it difficult for clinicians to review events and changes in symptoms over time. In this study, we expanded the summary output by presenting a visualization of the patient’s journey to facilitate communication between patients and medical staff. Referring to the widespread use of LLM for summarization, we compared GPT-4.1 and Gemini-2.5-pro, and used Gemini-3-pro-image-preview for visualization. Data was collected from DIPEx-Japan, then the quality of the summaries was evaluated quantitatively and the visualizations qualitatively. Quantitative evaluation using BLEU and ROUGE metrics showed that Gemini-2.5-pro achieved higher summary scores than GPT-4.1, and Japanese summaries scored higher than English ones. Conversely, English performed better than Japanese in temporal expression extraction using precision, recall, and F1 metrics, and the Gemini-2.5-pro model consistently outperformed GPT-4.1. In qualitative evaluation using the pairwise method, the timetable-based model was far superior with an overall win rate of 0.865 in Japanese and 0.969 in English compared to the baseline.

Medical research is highly specialized, making it difficult for patients and general readers to understand recent findings.Traditionally, text simplification, replacing technical terms with more accessible expressions, has been employed. However, this approach alone is limited in addressing a lack of background knowledge and often results in the loss of important information.Therefore, this study defines “rewriting for non-experts” as a rewriting process that, in addition to simplification, supplements essential background knowledge such as the significance of the research and reasons it is needed and proposes a method for implementing this process using large language models (LLMs).To verify the effectiveness of the proposed approach, a quantitative evaluation using automatic metrics was conducted. The results showed that the method combining the guidelines for human text creation with few-shot examples of reference texts achieved the highest scores.The expansion of the guidelines is planned as part of future work to enable the rewriting of scientific and technological information in a form that is accessible to a broader audience.

The rapid expansion of biomedical literature makes manual identification of novel drug-disease relationships increasingly difficult. Existing approaches have leveraged LLMs to mine abstracts or construct knowledge graphs for drug repurposing. There are two key limitations: finite context windows for capturing macro-level research trends, and single-pass black-box pipelines make it difficult to verify outputs. This paper proposes a pipeline for discovering new drug targets by combining disease and drug research trends using Large Language Models (LLMs). Our method extracts PICO components from PubMed abstracts, normalizing the Population and Intervention Component to ICD and ATC codes, respectively. A temporal frequency delta matrix is constructed to capture publication count shifts across 2013 to 2022, then used to discover novel drug areas. Compared with the abstract-based baseline, our approach showed qualitative signs of generating combinations that were more closely aligned with observed research trends and, in some cases, more clinically plausible. These findings suggest the potential usefulness of structured trend information for LLM-based exploration, although the differences between the two methods were limited and the results remain preliminary. Future work will focus on validating the consistency and reliability of these candidates.

User-generated data sources have gained significance in uncovering Adverse Drug Reactions (ADRs), with an increasing number of discussions occurring in the digital world. However, the existing clinical corpora predominantly revolve around scientific articles in English. This work presents a multilingual corpus of texts concerning ADRs gathered from diverse sources, including patient fora, social media, and clinical reports in German, French, and Japanese. Our corpus contains annotations covering 12 entity types, four attribute types, and 13 relation types. It contributes to the development of real-world multilingual language models for healthcare. We provide statistics to highlight certain challenges associated with the corpus and conduct preliminary experiments resulting in strong baselines for extracting entities and relations between these entities, both within and across languages.

Since medical text cannot be shared easily due to privacy concerns, synthetic data bears much potential for natural language processing applications. In the context of social media and user-generated messages about drug intake and adverse drug effects, this work presents different methods to examine the authenticity of synthetic text. We conclude that the generated tweets are untraceable and show enough authenticity from the medical point of view to be used as a replacement for a real Twitter corpus. However, original data might still be the preferred choice as they contain much more diversity.