Datasets:
text stringlengths 0 7.4k |
|---|
Abstract |
Climate change is intensifying hazards that threaten population health, including those that undermine hygiene behaviours, services, and infrastructure. While water and sanitation feature in climate–health agendas, hygiene, and the behaviours and enabling conditions that interrupt disease transmission, remains under-sp... |
Introduction |
Hygiene encompasses a diverse set of behaviours aimed at reducing the risk of disease transmission, spanning domestic and institutional settings. Modifying these behaviours can reduce the multiple exposure routes important for disease transmission through hands, food, and surfaces. Inadequate hygiene remains a major co... |
Climate change poses an escalating threat to population health. Over half of known human pathogenic diseases are aggravated by climatic factors, due to the alteration of behaviour, range, and longevity of vectors, hosts and reservoirs, and increased propagation of viruses, bacteria, and protozoa. As these disease press... |
However, the same climatic changes that heighten the need for hygiene also undermine the infrastructure, resources and services that enable it. Slow-onset events, such as drought and land degradation, and acute shocks, such as floods and cyclones, are undermining water and sanitation systems and disrupting critical sup... |
Climate hazards threaten all aspects of hygiene. Shifts in water resources and service reliability can trigger coping mechanisms that have direct implications for hygiene and health. Reduced water availability often forces households to prioritise water use for drinking and cooking over hygiene needs, reducing handwash... |
Alongside these slow-onset challenges, extreme weather events, such as floods and storms, can also reduce household access to water by damaging infrastructure. This can drive behaviours such as reduced hygiene and washing or laundering in contaminated waters. Menstruating women and girls face specific challenges during... |
Food hygiene is similarly threatened. Changes in ambient temperature, precipitation, and humidity affect food safety during storage and preparation, facilitating the growth of and spread of foodborne pathogens and toxins. Ocean warming, climate-driven acidification, and shifts in salinity and precipitation alter the bi... |
Research on the role of climate determinants in population health is increasing, supported by national adaptation plans and strategic research agendas that emphasise actions to reduce the impacts of climate change. Recent frameworks, including the National Academies of Medicine’s research agenda, the European Commissio... |
Hygiene does feature in a limited number of climate adaptation strategies and research agendas. In 2024, WHO and UNICEF released a joint review identifying new indicators to enhance national and global monitoring of climate-resilient WASH, including monitoring of hygiene-related infrastructure and behavioural elements.... |
Understanding the interplay between climate hazards, hygiene practices, and health outcomes is challenging due to the complex, multi-sectoral, and context-specific pathways involved. Factors such as social norms, poor infrastructure resilience, service delivery models, and the lack of prioritisation of adaptation under... |
This study sought to identify consensus-based research priorities at the intersection of climate change, hygiene, health, and well-being. By systematically identifying and ranking key research gaps through global expert engagement, the aim is to guide future research investments, inform policy and support programming t... |
Methods |
Ethics statement |
Ethical approval was received from the LSHTM Research Ethics Committee (No. 30443) for the survey and interviews. Before KIIs, respondents received a respondent information sheet and signed informed consent forms, which outlined confidentiality and risk of the procedure, as well as the voluntary nature of participation... |
Study design |
To define research priorities on climate change, hygiene, and health, we used a consultative approach based on the Child Health and Nutrition Research Initiative (CHNRI) method to identify research questions and prioritise them in a transparent, consultative, comprehensive and replicable manner. The CHNRI methodology h... |
Step 1. Question identification |
The scope of the research questions and research prioritisation exercise was agreed upon by the study team at the outset (Table 1). |
An initial set of potential questions was identified through a rapid scoping review and key informant interviews. Full details on these methods are provided in Tables A and B in S1 Appendix. A total of 116 peer-reviewed manuscripts and 55 key informant interviews formed this initial dataset. The study team reviewed man... |
The study team screened, de-duplicated and refined questions to ensure relevance to study aims. Questions were organised to address climate impact drivers and specific climate hazards following the Intergovernmental Panel on Climate Change (IPCC) framework. Climate impact drivers include changing precipitation patterns... |
We further organised the research questions by three broad hygiene domains, drawing on definitions of behaviour from the Capability, Opportunity and Motivation Behaviour (COM-B) model. We defined personal hygiene as behaviours focused on maintaining the cleanliness of the body to support health and well-being. Examples... |
Step 2. Question refinement |
Question identification (Step 1) resulted in a total of 141 research questions. These questions were reviewed by six technical advisors from academic institutions (see Acknowledgements). The advisors assessed the questions for relevance to climate change, hygiene, and health; checked for duplication; and provided edito... |
The study team classified these 57 questions according to definitions from the “4Ds framework” of description, delivery, development, or discovery (Table D in S1 Appendix). Description includes research to assess the existing situation. Delivery refers to research to evaluate existing interventions. Development describ... |
Step 3: Scoring and prioritisation |
To allow a ranking of the 57 questions refined (Step 2), we first established the criteria for scoring. The original CHNRI methodology includes 15 possible criteria for scoring research questions (Table C in S1 Appendix). These were shared with four members of the technical advisory group, who were then asked to alloca... |
Relevancy (criteria weight = 0.88): Will the research answer relevant evidence gaps? |
Answerability (criteria weight = 0.84): Can the research be feasibly completed and answered between now and 2035? |
Impact (criteria weight = 0.80): Is the proposed research likely to improve population health and well-being? |
Potential for translation (criteria weight = 0.80): Will the research generate knowledge that can be translated into feasible interventions? |
An online survey was developed using Qualtrics (Provo, Utah, USA). A targeted dissemination approach was taken, sharing the survey through existing mailing lists, professional networks, participant re-sharing, and social media platforms were used to make the survey widely available. |
Respondents were asked to score each question by the four priority criteria - relevancy, answerability, impact, and potential for translation. They indicated their assessment by selecting “Yes” (allocated 1 point), “Maybe” (0.5 points), “No” (0 points), or “Not my Area of Expertise” (excluded from the scoring). The sur... |
For each research question, the research priority score (RPS) was calculated (Tables E and F in S1 Appendix). Scores were converted into a scaled and weighted RPS ranging from 0 to 100%. To support decision-making at different levels, results were also stratified by WHO region and by organisation affiliation, allowing ... |
Results |
A total of 141 people from 46 countries completed the online survey (Fig 1), representing a diverse cross-section of stakeholders, including academic institutions, international and local non-governmental organisations (NGOs), multilateral agencies, and government bodies, and reflecting a broad set of geographic and di... |
The top 20 highest-ranked research questions, shown in Table 2, achieved consistently high weighted RPS ranging from 75–100%, indicating high consensus on their importance. These questions spanned the hygiene domains and reflected a range of climate-impact drivers and hazards, including water scarcity, extreme weather ... |
The top-ranked questions revealed a clear emphasis on building the foundational evidence base at the intersection of climate change, hygiene, and health. A substantial majority (75%) of the top 20 questions focused on descriptive research, aiming to better characterise the burden of infectious diseases linked to climat... |
One fifth (20%) of the top-ranked questions related to delivery, focusing on how to provide practical support measures and preparedness messaging to protect health during climate hazards. Examples included identifying effective hygiene promotion actions before or during extreme weather events (Q1, Q11), tailoring suppo... |
Across the top 20, climate hazards were a dominant focus, appearing in 11 questions. Many sought to understand immediate disruptions from floods, cyclones, and other extreme weather events on hygiene behaviours; the cascading and long-term impacts from damaged infrastructure; and the impact of chronic service disruptio... |
The rank order list of all 57 research questions and questions stratified by organisational affiliation or region of focus can be found in the S3 Appendix and S4 Appendix. These stratified rankings indicate some variation by organisation type and region, potentially reflecting differences in institutional roles, mandat... |
Discussion |
This work complements and strengthens other global initiatives to identify evidence needs on the links between climate and health. By explicitly focusing on hygiene, a domain often underrepresented in climate-health discussions, this exercise fills a critical gap and provides actionable priorities to guide investments ... |
The findings underline the importance of multidisciplinary and cross-sector collaboration. Climate change, hygiene, and health are interlinked through complex pathways that span environmental systems, infrastructure, human behaviour, and social determinants. Addressing the identified research gaps will require integrat... |
The strong leaning toward descriptive research underscores that the field remains at a stage where robust empirical data on exposure pathways, disease burdens, and behavioural responses are urgently needed. This foundational understanding appears to be viewed by experts as essential before effective interventions and l... |
Equity concerns were evident across the priorities. Several top-ranked questions explicitly addressed the compounded vulnerabilities faced by people with disabilities, women and girls managing menstrual hygiene, and communities confronting multiple overlapping climate hazards. This aligns with broader calls in climate ... |
Consistent with patterns observed in other climate-health research agendas [49,50,53–55], our findings show a stronger priority given to climate-related hazards, such as extreme weather events, than on slow-onset changes such as chronic heat exposure or progressive water salinisation. In our prioritisation, 11 of the t... |
The research questions that emerged from this process are necessarily broad and require further specification and contextualization before they are readily translated into research protocols. It is important to note that our research prioritisation process did not differentiate between research questions that reflect a... |
Integration with global policy and existing literature |
End of preview. Expand in Data Studio
Fresh10
Fresh10 is a tiny, document-bounded perplexity diagnostic for language-model training experiments. It contains four recent English documents in each of ten fixed domains: science, medicine, software, history, fiction, government, legal, economics, climate, and education. Every source was published on or after 2025-05-01 and the corpus was retrieved on 2026-08-12.
Fresh10 is deliberately too small to be a general benchmark. It is intended to show whether an optimization that improves in-distribution validation also behaves sensibly on several recent, heterogeneous domains. Per-domain results and their unweighted macro-average should be reported; this diagnostic should not be used as a training corpus or qualification threshold.
Token and shard contract
The ten files under shards/ use the llm.c GPT-2 v1 format: a 1,024-byte
little-endian header (magic 20240520, version 1, token count), followed by
little-endian uint16 token IDs. Tokenization is GPT-2 via tiktoken 0.11.0,
with vocabulary size 50,257 and EOT token 50,256.
Each document contributes one unscored EOT context token followed by exactly
2,048 scored tokens. Therefore each domain shard stores 8,196 tokens and scores
8,192 targets; the full diagnostic scores 81,920 targets. Evaluators must use
the spans in artifact.json and must not score a transition across document
boundaries. A fixed 2,048-token prefix can end mid-sentence.
artifact.json binds every shard by byte count, SHA-256, token count, document
span, and source metadata. It intentionally does not contain its own Hosting
revision: consumers pin the immutable repository commit after publication.
documents.json records the cleaned-text and raw-source provenance. Raw and
canonical-text hashes support provenance and reconstruction audits; the shard
SHA-256 values in artifact.json are the binding evaluation artifacts.
Cleaning and quality review
Canonical UTF-8 text for human review is included under texts/. The first
scored window of all 40 documents was inspected. Deterministic cleaners remove
site navigation and UI; PLOS metadata, references, citation markers, figures,
tables and captions; CRS small-print notes, repeated headers/footers, visual
matter and sources; Federal Reserve HTML table bodies, figures and widgets; and
SCP page chrome, footnote references and hidden UI. Ordinary prose references
to a table or figure are retained. Supreme Court opinions are extracted
page-by-page, and software release documents are normalized without assigning a
snapshot date to older content.
Licenses and attribution
This is a mixed-license collection and has no single replacement license. Reuse each component only under its source terms:
- PLOS articles are adapted under CC BY 4.0.
- SCP Wiki works and their cleaned adaptations are under CC BY-SA 3.0; retain creator attribution, a license link, and ShareAlike obligations.
- Software documents retain their Apache-2.0, PSF-2.0, BSD-3-Clause, or PostgreSQL license, as listed below.
- U.S. federal government works are identified as public domain in the United States. Other jurisdictions may differ.
The following is the complete creator/source attribution for the 40 documents.
Machine-readable titles, publishers, dates, URLs, licenses, extraction notes,
and hashes are preserved in documents.json and artifact.json.
| Domain | Work | Creator(s) | Terms | Published |
|---|---|---|---|---|
| science | A high-resolution phenotypic screen identifies regulators of CNS axon diameter in zebrafish | Maria A. Eichel-Vogel; Daniel Y. H. Soong; Melody N. Sequeira; Maya L. A. Ibenfeldt; Katy L. H. Marshall-Phelps; Jenea M. Bin; David A. Lyons | CC BY 4.0 | 2026-07-17 |
| science | What will be the future of computational biology for macromolecules in the era of AI? | Arne Elofsson; Nir Ben-Tal | CC BY 4.0 | 2026-07-08 |
| science | Mind the gap: An embedding guide to safely travel in sequence space | Adam Wu; Jakub Lála; Quentin Trolliet; Abhinav Rajendran; Stefano Angioletti-Uberti | CC BY 4.0 | 2026-07-09 |
| science | Extracting host-specific developmental signatures from longitudinal microbiome data | Balázs Erdős; Christos Chatzis; Jonathan Thorsen; Jakob Stokholm; Age K. Smilde; Morten A. Rasmussen; Evrim Acar | CC BY 4.0 | 2026-07-06 |
| medicine | Prediction of hospitalisation in young children with pneumonia in Malawi: A machine learning-based approach | Patrick Staunton; Mohammad Adib Makrooni; Master Chisale; Billy Nyambolo; Joseph Wu; Damien McCarthy; Mark Ledwidge; Yasir Bin Nisar; Chris Watson; Balwani Mbakaya; Cathal Seoighe; Joe Gallagher | CC BY 4.0 | 2026-06-09 |
| medicine | Multi-omics biomarkers of endothelial dysregulation preceding chronic lung allograft dysfunction | Giulia Iacono; Christina Begka; Bailey Cardwell; Carmel Daunt; Roxanne Chatzis; Celine Pattaroni; Alana Butler; Matthew Macowan; Bronwyn Levvey; Gregory I. Snell; Glen P. Westall; Benjamin J. Marsland | CC BY 4.0 | 2026-06-23 |
| medicine | Cardiovascular outcomes and safety associated with statin therapy for primary prevention in older adults with type 2 diabetes | Linda Chan; Wanchun Xu; Esther W. Y. Chan; Eric Yuk Fai Wan | CC BY 4.0 | 2026-06-24 |
| medicine | Plasma proteomic signatures of early retinal neurodegeneration in diabetes: a multi-cohort study | Huangdong Li; Ziyu Zhu; Shaopeng Yang; Weijing Cheng; Shaoying Tan; Zhuoyao Xin; Lei Zhang; Zhuoting Zhu; Shida Chen; Wenyong Huang; Wei Wang | CC BY 4.0 | 2026-06-02 |
| software | JAX change log through 0.11.0 | JAX project contributors | Apache-2.0 | 2026-07-16 |
| software | Python 3.14.6 changelog | Python Software Foundation contributors | PSF-2.0 | 2026-06-10 |
| software | NumPy 2.4.0 release notes | NumPy developers | BSD-3-Clause | 2025-12-20 |
| software | PostgreSQL 18 release notes | PostgreSQL Global Development Group | PostgreSQL License | 2025-09-25 |
| history | Dating ancient manuscripts using radiocarbon and AI-based writing style analysis | Mladen Popović; Maruf A. Dhali; Lambert Schomaker; Johannes van der Plicht; Kaare Lund Rasmussen; Jacopo La Nasa; Ilaria Degano; Maria Perla Colombini; Eibert Tigchelaar | CC BY 4.0 | 2025-06-04 |
| history | The Minoan Thera eruption predates Pharaoh Ahmose: Radiocarbon dating of Egyptian 17th to early 18th Dynasty museum objects | Hendrik J. Bruins; Johannes van der Plicht | CC BY 4.0 | 2025-09-10 |
| history | Amorphous carbonized objects and their contribution to reconstructing ancient Mesoamerican cuisine | Clarissa Cagnato; Nawa Sugiyama; Laura Longo; Alessandro Bonetto; Matteo Parisatto; Elena Longo; Marko Prasek; Giuliana Tromba; Antonio Marcomini; Elena Badetti | CC BY 4.0 | 2025-11-19 |
| history | 3000 yr-old patterns of mobile pastoralism revealed by multiple isotopes and radiocarbon dating of ancient horses from the Mongolian Altai | Antoine Zazzo; Maël Le Corre; Nicolas Lazzerini; Charlotte Marchina; Noost Bayarkhuu; Vincent Bernard; Mathilde Cervel; Denis Fiorillo; Dominique Joly; Michel Lemoine; Philippe Telouk; François Thil; Tsagaan Turbat; Vincent Balter; Aurélie Coulon; Sébastien Lepetz | CC BY 4.0 | 2025-05-07 |
| fiction | GARDEN OF THE GODS | sprawlingstar | CC BY-SA 3.0 | 2026-01-01 |
| fiction | SCP-3930 Does Not Exist | badooga | CC BY-SA 3.0 | 2026-02-06 |
| fiction | MY COWORKER IS FUCKING SATAN | FigaroFigaroFigaro | CC BY-SA 3.0 | 2026-02-05 |
| fiction | Pilares | Din-Bidor | CC BY-SA 3.0 | 2025-11-01 |
| government | Congressional Commissions: Overview and Considerations for Congress | Congressional Research Service | U.S. government work | 2026-01-21 |
| government | Federal Reserve: Policy Issues in the 119th Congress | Congressional Research Service | U.S. government work | 2026-01-14 |
| government | The Universal Service Fund and Related FCC Broadband Programs: Overview and Considerations for Congress | Congressional Research Service | U.S. government work | 2026-01-21 |
| government | Federal Research and Development (R&D) Funding: FY2026 | Congressional Research Service | U.S. government work | 2025-12-23 |
| legal | West Virginia v. B. P. J. | Supreme Court of the United States | Public domain (17 U.S.C. § 105) | 2026-06-30 |
| legal | National Republican Senatorial Committee v. Federal Election Commission | Supreme Court of the United States | Public domain (17 U.S.C. § 105) | 2026-06-30 |
| legal | Trump v. Barbara | Supreme Court of the United States | Public domain (17 U.S.C. § 105) | 2026-06-30 |
| legal | Watson v. Republican National Committee | Supreme Court of the United States | Public domain (17 U.S.C. § 105) | 2026-06-29 |
| economics | Monetary Policy Report – July 2026, recent economic and financial developments | Board of Governors of the Federal Reserve System | Public domain (17 U.S.C. § 105) | 2026-07-10 |
| economics | Funding Risks — Financial Stability Report, May 2026 | Board of Governors of the Federal Reserve System | Public domain (17 U.S.C. § 105) | 2026-05-08 |
| economics | Employment and Job Quality — Economic Well-Being of U.S. Households in 2025 | Board of Governors of the Federal Reserve System | Public domain (17 U.S.C. § 105) | 2026-05-13 |
| economics | Supervisory Developments — Supervision and Regulation Report, June 2026 | Board of Governors of the Federal Reserve System | Public domain (17 U.S.C. § 105) | 2026-06-11 |
| climate | Urban thermal inequity: Advancing thermal justice transitions for vulnerable low-income populations | Konstantina Vasilakopoulou; Ansar Khan; Priyadarsini Rajagopalan; Samantha Wijewardane; Matthaios Santamouris | CC BY 4.0 | 2026-06-18 |
| climate | Climate change, hygiene, and health: A research roadmap for climate adaptation | Jessica Gerard; Lauren D’Mello-Guyett; Jane Falconer; Camille Heylen; Sari Kovats; Robert Dreibelbis | CC BY 4.0 | 2026-05-11 |
| climate | A post-AR6 update on observed and projected climate change in India | Chirag Dhara; Aditi Deshpande; Mathew Koll Roxy; Padmini Dalpadado; Mandira Singh Shrestha | CC BY 4.0 | 2025-11-20 |
| climate | Comparison of multidecadal variability in climate reanalyses and global models | Andrew Westgate; Sergey Kravtsov | CC BY 4.0 | 2025-10-07 |
| education | A dual-path framework for enhancing student engagement and learning outcomes in sports education | Guifeng Zheng; Yanting Wang; Juan Du | CC BY 4.0 | 2026-03-25 |
| education | Cognitive load and pedagogical tension in multi-platform online learning: Evidence from Chinese higher education | Xiaoqin Qiu; Nengsheng Qiu | CC BY 4.0 | 2026-04-21 |
| education | Teacher support for generative AI use and design learning outcomes in fashion design education | Zhiyi Zhang | CC BY 4.0 | 2026-05-06 |
| education | What factors enhance students' achievement? A machine learning and interpretable methods approach | Hui Mao; Ribesh Khanal; ChengZhang Qu; HuaFeng Kong; TingYao Jiang | CC BY 4.0 | 2025-05-16 |
Reconstructing the artifacts
The published build was produced with Python 3.10.12, tiktoken 0.11.0, requests 2.32.5, beautifulsoup4 4.14.2, pypdf 6.1.1, and pdftohtml/Poppler 22.02.0. CRS PDF geometry extraction is Poppler-version sensitive; exact text and artifact locks turn any drift into a hard failure.
From the repository root, with Poppler 22.02.0 available:
python3 builder/fetch_source_pack.py
python3 builder/clean_source_pack.py
uv run --python 3.10.12 \
--with tiktoken==0.11.0 \
--with requests==2.32.5 \
--with beautifulsoup4==4.14.2 \
--with pypdf==6.1.1 \
builder/build_fresh10.py
The first command accepts only the 20 upstream files whose bytes match
builder/source_pack.json; four frozen CC BY-SA SCP HTML snapshots are included
under builder/sources/ because the live pages can drift. The build validates
all 40 raw and canonical-text locks in documents.json, reconstructs all ten
shards in a sibling staging directory, compares the complete result to
artifact.json, and replaces build/ only after every check passes. Mutable
upstream bytes are never silently substituted.
- Downloads last month
- 57