Holistix: A Dataset for Holistic Wellness Dimensions Analysis in Mental Health Narratives
Heba Shakeel, Tanvir Ahmad, Chandni Saxena · 2025
Read Holistix: A Dataset for Holistic Wellness Dimensions Analysis in Mental Health Narratives by Heba Shakeel, Tanvir Ahmad, Chandni Saxena (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-09565.txt.
Introduction
Heba Shakeel, Tanvir Ahmad, and Chandni Saxena introduce Holistix, a dataset on wellness in social media posts, in this 2025 paper. Mental health writing online reflects many sides of well-being at once. The dataset sorts text into six wellness dimensions.
The six aspects are physical, emotional, social, intellectual, spiritual, and vocational health. Domain experts guided the annotation rules that label text spans. The team tested both classic machine learning and transformer models on the sorting task. Scores used precision, recall, and F1 across ten-fold checks. Explanation tools were added so model choices stay transparent. The data support local wellness checks and early help plans.
Catalog record
| CREATOR | Heba Shakeel, Tanvir Ahmad, Chandni Saxena |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 28 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-09565
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Holistix: A Dataset for Holistic Wellness Dimensions Analysis in Mental Health Narratives Heba Shakeel, Tanvir Ahmad, Chandni Saxena Abstract We introduce a dataset for classifying wellness dimensions in social media user posts, covering six key aspects: physical, emotional, social, intellectual, spiritual, and vocational. The dataset is designed to capture these dimensions in user-generated content, with a comprehensive annotation framework developed under the guidance of domain experts. This framework also includes labeling text spans from these posts to provide explanations that highlight the corresponding wellness aspects. We evaluate both traditional machine learning models and advanced transformer-based models for this multi-class classification task, with performance assessed using precision, recall, and F1-score, averaged over 10-fold cross-validation. Post-hoc explanations are applied to ensure the transparency and interpretability of model decisions. The proposed dataset contributes to region-specific wellness assessments in social media and paves the way for personalized well-being evaluations and early intervention strategies in mental health. We adhere to ethical considerations for constructing and releasing our experiments and dataset publicly on Github 1 1 1 https://github.com/HebaShakeel/holistix . I Introduction Mental health disorders have become a critical global health problem, with more than a billion people worldwide affected by mental, neurological, and substance use disorders [ 1 ] . Depression alone impacts approximately 280 million people and stands as a leading cause of disability globally [ 2 ] . Despite the prevalence of these conditions, access to timely and effective mental health care remains a significant challenge, often due to stigma, resource limitations, and a shortage of mental health professionals [ 3 ] . In parallel, the advent of social media has transformed communication, with platforms like Facebook, Twitter, and Instagram becoming integral to daily life. While these platforms offer avenues for connection and information sharing, excessive use has been linked to negative mental health outcomes, including increased anxiety and depression [ 4 ] . Conversely, social media also provides a unique opportunity for mental health assessment and intervention [ 5 ] . Individuals often share their thoughts and emotions online, creating a rich data source that, with appropriate analysis, can offer an understanding of population-level mental health trends. Fig. 1: Overview of the problem formulation showing the identification of wellness dimensions in user posts. Studies have demonstrated the potential of AI and NLP techniques to analyze social media content, enabling monitoring of mental health indicators across diverse populations [ 6 , 7 ] . By leveraging the vast amount of user-generated content, researchers and practitioners can develop innovative, non-intrusive methods for early detection, continuous assessments, and personalized interventions [ 8 ] . Despite the potential of AI and NLP techniques to analyze social media content for mental health monitoring, several challenges persist. One significant issue is the scarcity of high-quality, annotated datasets that capture the multifaceted nature of mental health expressions across diverse populations [ 9 ] . Existing datasets often lack the granularity required to identify specific wellness dimensions, such as emotional, social, and psychological well-being, which are crucial for comprehensive mental health assessments. To address this gap, we introduce a fine-grained approach to mental health assessment that frames the task as wellness concept classification accompanied by explanatory text span annotations. This involves creating a novel, population-specific ‘‘Holistix” dataset focusing on the Australian population, with data collected from users reporting on a mental health forum 2 2 2 https://www.beyondblue.org.au . Our dataset comprises 1420 instances totaling 37082 words, annotated by experts following a structured annotation scheme grounded in Halbert L. Dunn’s theory of wellness dimensions. Fig. 1 illustrates the potential for tailoring mental health assessments by analyzing human writings through the lens of wellness dimensions, thereby moving beyond basic evaluations. Notably, while previous studies, such as [ 10 ] , utilized Reddit data and focused on four of Dunn’s wellness dimensions, our research advances this by encompassing all six dimensions: physical, emotional, intellectual, spiritual, social, and vocational aspects [ 11 ] . We refer to the six wellness dimensions as ‘aspects’ to maintain consistency in terminology with recent works [ 10 , 12 ] . This comprehensive approach provides a more holistic understanding of mental health as expressed in social media contexts. We fine-tune baseline transformer-based models and use traditional machine learning techniques for multi-class wellness dimension classification. The proposed dataset contributes to region-specific wellness assessments in social media and paves the way for personalized well-being and early mental health intervention. II Holistix Dataset The utilization of social media data has become pivotal in understanding mental health trends and behaviors. Platforms like Facebook [ 13 ] , Twitter [ 14 ] , Reddit [ 15 ] , and region-specific platforms like Weibo [ 16 ] offer rich, user-generated content that researchers analyze to gain insights into public sentiment and mental health issues. Although some data resources are publicly available, many datasets remain inaccessible due to concerns over user privacy and adherence to ethical standards. Reddit, with its diverse and anonymous user base, provides valuable data for mental health research [ 17 , 18 , 15 , 19 , 10 ] , while platforms like Weibo [ 16 ] (a Chinese microblog), offer localized insights into region-specific mental health discussions. Comparative analyses of global data from Reddit and country-specific data from Reddit [ 20 ] reveal significant cross-cultural differences in mental health expressions, underscoring the need for culturally sensitive research and interventions. Despite existing studies, there remains a critical need for more country-specific datasets to capture cultural, social, and linguistic nuances influencing mental health. To address this gap, we have curated the Holistix dataset from Australia’s Beyond Blue [ 21 ] mental health forums, reflecting the unique challenges faced by the Australian population. This data will help develop targeted, culturally appropriate interventions and enrich mental health research with localized perspectives. II-A Dataset Creation Beyond Blue [ 21 ] , one of Australia’s most reputable mental health organizations, provides a secure and anonymous platform for individuals to access information, seek advice, and share experiences. The anonymity fosters open discussions, making it a more viable resource for mental health research compared to conventional social media platforms. We extracted 2,000 raw posts from Beyond Blue’s online forums using the Python library- BeautifulSoup 3 3 3 https://pypi.org/project/beautifulsoup4/ . The data was sourced from key discussion categories, including Anxiety , Depression , PTSD and Trauma , Suicidal Thoughts and Self-Harm , Relationship and Family Issues , Supporting Friends and Family , and Grief and Loss . To ensure user privacy, only the textual content and its corresponding category were retained. To enhance data quality, we performed rigorous preprocessing by removing irrelevant, empty, and duplicate posts. Posts that were excessively long or off-topic were also filtered out, refining the dataset to 1,420 posts specifically focused on mental distress. II-B Annotation Framework For accurate classification, domain experts that include a social NLP researcher and a senior clinical psychologist who developed structured annotation guidelines. They collaboratively defined clear class indicators, annotation schemes, and perplexity criteria to categorize posts based on underlying wellness dimensions, ensuring consistency and reliability in the labeling process. II-B 1 Wellness Dimensions Building upon the foundational work of Halbert L. Dunn and Dr. Bill Hettler, our research focuses on the intersection of these wellness dimensions and mental health within