Health Nexus: Digital Health and Medical AI supports responsible data management, transparent reporting, reproducibility, validation, and the appropriate reuse of research outputs. Authors are expected to make the data and materials underlying their published findings available to the greatest extent permitted by ethical approval, informed consent, privacy and confidentiality obligations, applicable law, contractual terms, intellectual-property rights, institutional policy, security considerations, and legitimate commercial restrictions.

All research articles must include a clear and accurate Data Availability Statement, regardless of whether the supporting data are openly available, available under controlled conditions, available upon reasonable request, or cannot be shared.

A statement such as “data are available” without identifying the location, access conditions, responsible party, or applicable restrictions is insufficient.

---

1. Purpose of the Policy

The purposes of this policy are to:

• Support verification and reproducibility of published findings;
• Enable responsible reuse of valuable research data;
• Promote transparency concerning the evidence underlying published claims;
• Protect research participants, patients, institutions, communities, and data providers;
• Clarify the circumstances in which data can be accessed;
• Encourage appropriate documentation, preservation, citation, and attribution of research data;
• Reduce avoidable duplication of research effort;
• Support responsible development and evaluation of digital-health and medical-AI systems;
• Maintain the accuracy and integrity of the scholarly record.

Data sharing is not an absolute requirement to make every research record publicly downloadable. The journal requires transparent disclosure and expects the maximum appropriate level of access consistent with ethical, legal, scientific, technical, and security obligations.

---

2. Scope of the Policy

This policy applies, where relevant, to:

• Quantitative and qualitative research data;
• Clinical-trial and observational-study data;
• Electronic health-record and registry data;
• Medical images, biosignals, genomic data, and laboratory data;
• Survey responses and interview materials;
• Digital-health platform and wearable-device data;
• Training, validation, tuning, and test datasets used in artificial-intelligence research;
• Synthetic datasets and simulation outputs;
• Statistical analysis files and derived datasets;
• Research code, software, algorithms, model documentation, and model weights;
• Prompts, system instructions, evaluation rubrics, and generated outputs used in large language model research;
• Study protocols, statistical analysis plans, data dictionaries, questionnaires, annotation manuals, and other supporting materials;
• Supplementary files necessary to understand or verify the published work.

The policy applies to original research articles, clinical studies, diagnostic and prognostic studies, AI and machine-learning studies, implementation studies, methodological studies, technical and software articles, protocols, systematic reviews, meta-analyses, and other submissions that generate or analyse research data.

---

3. Mandatory Data Availability Statement

Every research article must contain a section titled Data Availability Statement.

The statement must indicate:

• Whether data were generated or analysed;
• Which data support the article’s principal findings;
• Whether the data are publicly available, available under controlled access, available upon request, or unavailable;
• The name of the repository or data custodian;
• The DOI, accession number, persistent identifier, or permanent repository link where applicable;
• The time at which the data will become available;
• The expected duration of availability;
• Any conditions or restrictions governing access;
• The identity or role of the party responsible for reviewing access requests;
• Whether supporting code, models, prompts, protocols, analysis plans, or other materials are available;
• The ethical, legal, privacy, security, contractual, or commercial basis for any limitation.

The statement must reflect the actual arrangements in place at the time of publication. Authors must not claim that data are available where no practical access mechanism exists.

---

4. Publicly Available Data

Where supporting data can be shared openly, authors should deposit them in a trusted repository that provides:

• A persistent identifier;
• A stable public landing page;
• Clear metadata;
• Version information;
• A stated licence or terms of use;
• Long-term preservation arrangements;
• Appropriate access and download functionality.

Suggested statement:

Where several datasets or repositories are used, each relevant dataset and identifier should be listed clearly.

---

5. Data Available under Controlled Access

Controlled access is appropriate where data cannot be made openly downloadable but may be shared with qualified applicants under defined safeguards.

The Data Availability Statement should identify:

• The data custodian or access committee;
• The eligibility requirements for applicants;
• The permitted research purposes;
• The application and review procedure;
• Required ethics approval or institutional authorisation;
• Required data-use, confidentiality, or security agreements;
• Any restrictions on onward sharing, linkage, publication, or commercial use;
• The expected review timeframe;
• Whether access fees apply;
• The period for which approved access will be granted.

Suggested statement:

Controlled access must not be described as open access.

---

6. Data Available upon Reasonable Request

The journal permits a reasonable-request model where repository deposition or open release is not practicable, provided that the arrangement is genuine, transparent, and sustainable.

A reasonable-request statement should explain:

• Who may request the data;
• What information must be included in the request;
• What types of analyses or purposes are eligible;
• Who will review the request;
• What ethical, legal, institutional, or contractual approvals are required;
• Whether a data-use agreement is required;
• Which data and documentation may be provided;
• Why the data have not been deposited in a public or controlled-access repository;
• How long the data are expected to remain available.

Suggested statement:

The phrase “available upon reasonable request” must not be used where the authors, institution, or sponsor do not intend to consider legitimate requests.

The journal may request evidence that the stated access procedure remains operational.

---

7. Restricted or Unavailable Data

Data may be withheld where sharing would conflict with legitimate obligations or create a material risk that cannot be mitigated adequately.

Acceptable reasons may include:

• Participant consent does not permit the proposed sharing or reuse;
• Data contain identifiable or re-identifiable health information;
• Public disclosure would create a substantial privacy, safety, or discrimination risk;
• Applicable law or regulation prohibits disclosure;
• An ethics committee or data custodian has imposed a valid restriction;
• Data are owned or controlled by a third party;
• A contractual data-use agreement prohibits redistribution;
• Data concern vulnerable, marginalised, Indigenous, or otherwise protected communities requiring additional governance;
• Disclosure would create a serious cybersecurity or infrastructure risk;
• The data contain confidential commercial, proprietary, or security-sensitive information;
• The authors do not possess lawful authority to redistribute the data.

A restriction must be explained specifically. Statements such as “data cannot be shared for confidentiality reasons” should identify the relevant category of restriction without disclosing confidential information.

Suggested statement:

Where no access can lawfully be provided:

---

8. Studies Using Third-Party Data

Authors who analyse data obtained from another institution, repository, government agency, healthcare provider, commercial platform, or data custodian must:

• Identify the original data source;
• Cite the dataset using its persistent identifier where available;
• Describe the procedure through which access was obtained;
• State the relevant licence or data-use conditions;
• Explain whether other researchers can obtain access through the same procedure;
• Avoid uploading or redistributing data they are not authorised to share;
• Distinguish clearly between source data and datasets created by the authors.

Suggested statement:

Where third-party data are publicly available, authors should provide the original repository record and persistent identifier rather than uploading an unauthorised duplicate.

---

9. No New Data Generated or Analysed

Articles that do not generate or analyse new research data should state this explicitly.

Suggested statement:

This statement may be appropriate for certain editorials, perspectives, conceptual articles, and narrative reviews. It should not be used for systematic reviews, meta-analyses, computational studies, or other work that creates extraction files, analytical datasets, code, or derived results.

---

10. Systematic Reviews and Meta-Analyses

Systematic reviews and meta-analyses should make available, where legally permissible:

• The complete search strategies;
• The study-screening record;
• The inclusion and exclusion decisions;
• The data-extraction form or extracted dataset;
• The risk-of-bias assessments;
• The statistical analysis code;
• The meta-analysis input files;
• Any evidence-grading materials;
• The protocol and registration information.

Copyrighted full-text articles obtained during the review should not be redistributed unless the authors possess authority to do so.

---

11. Clinical Trials

Manuscripts reporting clinical trials must contain a data-sharing statement that specifies:

• Whether de-identified individual participant data will be shared;
• Which individual participant data will be shared;
• Whether a data dictionary will be available;
• Whether the study protocol, statistical analysis plan, informed-consent form, clinical study report, or analytical code will be shared;
• When the materials will become available;
• How long they will remain available;
• Who may obtain access;
• For what purposes access may be granted;
• The application and approval mechanism;
• Whether a data-use agreement or other condition applies.

“Undecided” is not an acceptable clinical-trial data-sharing statement.

For trials beginning enrolment on or after the date required by the applicable clinical-trial registration standards, the data-sharing plan should also be included in the public trial-registration record.

If the data-sharing plan changes after registration, authors must:

• Update the registry record where possible;
• Explain the change in the submitted manuscript;
• Ensure that the published statement accurately reflects the final arrangement.

Example of a clinical-trial statement:

---

12. Human-Participant and Clinical Data

Open publication of an article does not authorise unrestricted release of participant-level data.

Before sharing human-participant data, authors and their institutions must consider:

• The original informed-consent language;
• The scope of ethics approval;
• Applicable health-information and data-protection law;
• Direct and indirect identifiers;
• The possibility of re-identification through data linkage;
• The sensitivity and rarity of the condition or population;
• Geographic, occupational, familial, genomic, biometric, and temporal identifiers;
• Risks to participants, families, communities, and institutions;
• Whether open, controlled, or secure-enclave access is appropriate;
• Whether additional consent or governance review is required.

Removing names and direct identifiers does not automatically make a dataset anonymous. Authors must assess the entire dataset and the availability of external information that could permit re-identification.

The journal will not require public disclosure of data where doing so would violate participant rights, ethics approval, applicable law, or legitimate confidentiality obligations.

---

13. Qualitative Research Data

Qualitative data may present substantial confidentiality and contextual-identification risks. Interview transcripts, field notes, audio recordings, videos, and detailed narratives must not be shared openly where participants may be identifiable or where consent does not permit such use.

Authors should consider sharing, where appropriate:

• De-identified extracts;
• Coding frameworks;
• Codebooks;
• Analytical memos;
• Interview guides;
• Theme-development records;
• Aggregated or redacted data;
• Controlled-access versions of transcripts.

Authors must explain the balance between reproducibility and participant confidentiality in the Data Availability Statement.

---

14. Artificial-Intelligence and Machine-Learning Data

Studies involving artificial intelligence, machine learning, deep learning, clinical prediction models, medical imaging, natural-language processing, or related technologies should identify the availability and governance of:

• Training datasets;
• Validation and tuning datasets;
• Independent test datasets;
• External-validation datasets;
• Labels, annotations, and reference standards;
• Data dictionaries and variable definitions;
• Preprocessing and feature-engineering procedures;
• Data-splitting logic;
• Code used to train and evaluate the model;
• Model architecture and hyperparameters;
• Model weights or checkpoints;
• Evaluation scripts and performance outputs;
• Subgroup and fairness analyses;
• Model cards, data sheets, or equivalent documentation.

Where datasets or models cannot be shared, authors must provide sufficient methodological detail to permit critical evaluation and must explain the applicable restriction.

Authors must distinguish clearly between:

• Publicly available data;
• Institutionally controlled data;
• Commercial or proprietary data;
• Data accessible only through an external application process;
• Data that cannot be redistributed by the authors;
• Synthetic data generated for the study.

A statement that a proprietary dataset or model is unavailable must not be used to omit information necessary to evaluate data provenance, study population, preprocessing, validation, bias, limitations, or conflicts of interest.

---

15. Generative AI and Large Language Model Studies

Where a study involves generative AI, large language models, foundation models, multimodal systems, chatbots, or autonomous agents, the availability statement should address, where applicable:

• System and user prompts;
• Prompt templates;
• System instructions;
• Model name and version;
• Date and method of access;
• Application programming interface settings;
• Temperature, token limits, and other generation parameters;
• Conversation histories or context windows;
• Retrieval documents or knowledge bases;
• Generated outputs;
• Human-evaluation rubrics;
• Annotator instructions;
• Scoring criteria and evaluation code;
• Repeated-query procedures;
• Records necessary to assess model drift or version change.

Where platform terms, privacy requirements, copyright restrictions, or technical limitations prevent full sharing, authors should archive and share the maximum amount of lawful documentation necessary to understand and verify the evaluation.

---

16. Synthetic Data

Synthetic data must be labelled explicitly and must not be represented as authentic participant, patient, clinical, observational, or experimental data.

Where synthetic data are shared, the deposited record should describe:

• The generation procedure;
• The source and governance of the original data;
• The model and version used;
• Generation parameters;
• Privacy and disclosure-risk assessment;
• Fidelity and utility evaluation;
• Known biases and limitations;
• Whether the synthetic data may be used for model training, validation, demonstration, or another purpose.

A synthetic dataset must not be described as anonymous or privacy-preserving without appropriate evidence.

---

17. Code, Software, Models, and Computational Materials

Authors are encouraged to make available the computational materials necessary to reproduce the published analysis, including:

• Source code;
• Analysis scripts;
• Software version information;
• Package and library dependencies;
• Configuration files;
• Random seeds;
• Model architecture;
• Hyperparameters;
• Model weights or checkpoints;
• Container or environment files;
• Testing and evaluation code;
• Instructions for reproducing the principal results.

Software and code should normally be deposited in a repository that supports release versioning and persistent archiving.

The licence applied to software or code should be stated separately. The CC BY 4.0 licence applied to the article may not be the most appropriate licence for functional software.

Where proprietary software is required, authors should identify the product, version, provider, and relevant settings and should explain whether the analysis can be reproduced using an alternative method.

---

18. Study Protocols and Analysis Plans

Authors should make available, where applicable:

• The original study protocol;
• Protocol amendments;
• The statistical analysis plan;
• Clinical-trial registration records;
• Data-management and sharing plans;
• Case-report forms;
• Survey instruments;
• Interview guides;
• Annotation manuals;
• Standard operating procedures necessary to understand the study.

Any material difference between the protocol, registry record, analysis plan, and published article must be explained.

---

19. Repository Selection

Authors should use a recognised disciplinary, institutional, governmental, or general-purpose repository appropriate to the type and sensitivity of the research output.

When selecting a repository, authors should consider whether it provides:

• A persistent identifier;
• Stable access and preservation;
• Version control;
• Clear metadata requirements;
• A defined licence or terms of access;
• Controlled-access mechanisms where required;
• Security appropriate to the sensitivity of the data;
• Support for the relevant file types and research discipline;
• A clear governance and sustainability model;
• A process for correcting, updating, or withdrawing a record.

Personal websites, temporary file-sharing services, private cloud folders, email attachments, or links without long-term preservation arrangements are not preferred as the sole method of data availability.

Where a discipline-specific repository is mandated by a funder, institution, registry, or law, authors should comply with that requirement.

---

20. Timing of Data Release

Supporting data should normally be available no later than the date on which the final Version of Record is published, unless:

• A justified embargo has been approved;
• A clinical-trial sharing plan specifies a later date;
• A repository or data custodian operates an established release schedule;
• Ethical, legal, contractual, or intellectual-property considerations require delayed access.

Any embargo must be stated in the Data Availability Statement and should identify:

• The reason for the embargo;
• The date on which access will begin;
• The material covered by the embargo;
• The access mechanism that will operate after the embargo expires.

Authors must not promise future data release without establishing a realistic preservation and access arrangement.

---

21. Duration of Availability

Data and supporting materials should remain available for a period appropriate to the discipline, research design, funder requirements, applicable law, repository policy, and institutional retention rules.

Where a finite access period applies, the Data Availability Statement must state:

• The date on which access begins;
• The date or condition on which access ends;
• The reason for the limitation;
• Whether the data will be transferred, archived, destroyed, or placed under different governance afterward.

Where possible, authors should select repositories that support long-term preservation beyond the active life of a research project or individual research account.

---

22. Metadata and Documentation

Shared data should be accompanied by sufficient documentation to permit interpretation and responsible reuse.

Documentation may include:

• Dataset title and description;
• Creator names and affiliations;
• Collection dates and locations;
• Study population and eligibility criteria;
• Variable names and definitions;
• Units and coding conventions;
• Missing-value codes;
• Data-cleaning and preprocessing procedures;
• Provenance and version history;
• File-format descriptions;
• Quality-control procedures;
• Known limitations;
• Ethical and legal access conditions;
• Recommended citation;
• Licence or terms of reuse.

Data without adequate documentation may be unusable or misleading and may not satisfy the journal’s transparency expectations.

---

23. File Formats and Version Control

Authors should use non-proprietary, widely supported, machine-readable formats where practical.

Deposited records should identify:

• The file format;
• The software required to open or analyse the file;
• The version of the dataset or software;
• The date of release;
• Material changes between versions;
• Whether the deposited version corresponds to the published analysis.

When data or code are updated after publication, earlier versions should remain identifiable where repository policy permits, and the article record should be corrected if the change affects the published findings.

---

24. Data Licensing and Terms of Reuse

Authors should identify a clear licence or terms of use for openly shared data and supporting materials.

The selected licence must be compatible with:

• Consent and ethics approval;
• Applicable privacy and data-protection law;
• Third-party rights;
• Repository requirements;
• Institutional and funder policies;
• The nature of the research output.

The article-level CC BY 4.0 licence does not automatically apply to external datasets, source code, software, model weights, questionnaires, or third-party materials.

Authors must not apply an open licence to material they do not have authority to license.

---

25. Data Citation

Datasets used or generated in the research should be cited as scholarly research outputs where a formal citation is available.

Dataset references must follow the PSG Author–Date Citation Format and should identify:

• The creator or responsible organisation;
• The year;
• The dataset title;
• The resource type;
• The repository;
• The version, where applicable;
• The DOI or persistent identifier.

Example:

Lee, Sungho. 2025. Clinical Dataset for Remote Patient Monitoring Research. Data set. Zenodo. https://doi.org/10.xxxx/xxxxx

In-text citation: (Lee 2025)

Authors conducting secondary analyses must cite the original dataset and acknowledge the researchers, institutions, participants, communities, and infrastructure responsible for its creation where appropriate.

---

26. Secondary Analysis and Reuse of Shared Data

Authors using data created by others must:

• Comply with the licence, consent, access agreement, and permitted uses;
• Cite the original dataset and related publications;
• Identify the dataset version used;
• State the date of access;
• Explain how the new analysis differs from previous analyses;
• Obtain any required ethics approval or exemption;
• Avoid attempting to re-identify participants;
• Respect restrictions on linkage, redistribution, commercial use, and onward access;
• Provide appropriate credit to data creators and custodians.

Access to shared data does not imply permission to use the data for any purpose.

---

27. Editorial and Peer-Review Access to Data

The journal may request access to data and supporting materials during editorial assessment, peer review, production, or post-publication investigation.

Requested materials may include:

• Underlying or minimally processed data;
• De-identified participant-level data;
• Statistical analysis files;
• Original images or signals;
• Code and software environments;
• Model outputs and evaluation records;
• Prompts and generated responses;
• Data dictionaries;
• Ethics or data-access documentation;
• Audit trails and version records.

Where sensitive materials cannot be transferred to the journal or reviewers, authors should propose a lawful and secure method of verification, such as:

• Controlled reviewer access;
• A secure research environment;
• Independent institutional verification;
• Review by an authorised specialist;
• Provision of aggregated, redacted, or synthetic validation materials.

Reviewers must treat all non-public data as confidential and must not retain, reuse, redistribute, or upload them to unauthorised systems.

---

28. Data Integrity and Accuracy

Authors are responsible for ensuring that deposited data and documentation:

• Correspond to the data used in the published analysis;
• Do not contain fabricated, falsified, substituted, or misleading records;
• Have not been selectively altered to support the published conclusions;
• Preserve relevant exclusions, missing-data information, and processing history;
• Do not contain unauthorised personal or confidential information;
• Are described accurately in the Data Availability Statement.

Depositing only selected records that create a misleading impression of the complete dataset may constitute a research-integrity concern.

The journal may seek independent technical, statistical, image, or data-integrity review where necessary.

---

29. Responsibilities of Authors and Data Custodians

Authors must identify who is responsible for:

• Maintaining the dataset;
• Reviewing access requests;
• Providing documentation;
• Correcting repository records;
• Responding to questions about the data;
• Ensuring continued compliance with ethical and legal obligations.

Corresponding authors should not promise access on behalf of an institution, sponsor, repository, or data custodian without confirming that they have authority to do so.

Where control of the data will transfer after the project ends, the long-term custodian should be identified.

---

30. Intellectual Property and Commercial Restrictions

The journal recognises that some research may involve patents, proprietary datasets, commercial software, licensing agreements, or confidential industrial information.

Commercial sensitivity alone does not justify withholding all information necessary to assess the validity of a study.

Authors must disclose:

• The identity of the relevant data or model owner;
• The nature of the restriction;
• Whether the authors had full access to the data;
• Whether the sponsor could restrict analysis or publication;
• Whether independent researchers can obtain access;
• What data, code, documentation, or aggregate information can be shared.

At least one accountable author should have sufficient access to the supporting data to verify the analysis and assume responsibility for the published findings.

---

31. Data-Sharing Costs

Authors may require applicants to cover reasonable, direct, and documented costs of preparing, transferring, or administering controlled access where such charges are permitted by the applicable institution or repository.

Charges must not be used to create an arbitrary or discriminatory barrier to legitimate scholarly access.

The Data Availability Statement should disclose any known access fee or cost-recovery requirement.

---

32. Changes to Data Availability after Publication

Authors must notify the journal promptly where:

• A repository link or DOI no longer functions;
• A dataset is withdrawn, restricted, replaced, or corrected;
• An access committee or data custodian changes;
• The stated access mechanism is no longer operational;
• A privacy, consent, security, or legal concern affects availability;
• A new version of the data materially affects the published analysis;
• The original Data Availability Statement is found to be inaccurate.

The journal may update the article metadata or publish a correction where a material change affects readers’ ability to access or interpret the supporting data.

Removal of data from a repository does not automatically remove the published article, but the journal will assess whether the article remains verifiable and reliable.

---

33. Repository Withdrawal or Removal

Data should not be removed from a repository solely to prevent legitimate scrutiny of published findings.

Removal or restriction may be appropriate where continued availability would:

• Violate law, consent, or privacy obligations;
• Create an unacceptable re-identification risk;
• Expose confidential or proprietary information unlawfully;
• Create a serious security or safety risk;
• Infringe third-party rights;
• Preserve materially incorrect or corrupted data without adequate warning.

Where data are removed, authors should retain a public repository record or tombstone notice explaining the general reason, unless prohibited by law or privacy obligations.

---

34. Non-Compliance

Potential non-compliance includes:

• Omission of a required Data Availability Statement;
• A statement that is materially false or misleading;
• Failure to deposit data promised before publication;
• Refusal to consider access requests despite stating that data are available;
• Providing data that do not correspond to the published analysis;
• Removing data to obstruct verification;
• Sharing personal or confidential information without authorisation;
• Failure to comply with participant consent or data-use conditions;
• Fabrication, falsification, or selective presentation of deposited data;
• Failure to disclose restrictions imposed by a sponsor or third party.

Depending on the seriousness and timing of the concern, the journal may:

• Request clarification or revision;
• Require repository deposition before acceptance;
• Pause peer review or production;
• Reject the manuscript;
• Withdraw an acceptance decision;
• Publish a correction or expression of concern;
• Retract an article whose findings cannot be considered reliable;
• Contact the authors’ institution, ethics committee, funder, repository, or data custodian;
• Take other proportionate action under the Publication Ethics policy.

A legitimate inability to share data is not misconduct where the restriction is disclosed accurately and the study remains capable of appropriate scholarly evaluation.

---

35. Suggested Data Availability Statements

Public Repository

Public Data and Code

Controlled Access

Reasonable Request

Third-Party Data

Sensitive Human Data

No Lawful Sharing Mechanism

No New Data

Data Included in the Article

Embargoed Data

---

36. Relationship to Other Journal Policies

This policy should be read together with the journal’s:

• Author Guidelines;
• Publication Ethics;
• AI and Research Integrity Policy;
• Peer Review Policy;
• Copyright and Licensing Policy;
• Citation and Reference Style;
• Corrections and Retractions Policy.

Where a funder, institution, ethics committee, registry, repository, law, or contractual agreement imposes a stricter requirement, authors must comply with that requirement and disclose its effect on data availability.

---

37. External Standards

The journal’s approach is informed by recognised standards concerning biomedical data sharing, clinical trials, research transparency, participant protection, and responsible data stewardship.

• ICMJE Clinical Trials and Data Sharing
• NIH Data Management and Sharing Policy
• NIH Guidance on Data Management and Sharing Plans

---

38. Policy Review

This policy may be revised to reflect developments in:

• Medical and health-data governance;
• Privacy and data-protection law;
• Clinical-trial requirements;
• Repository and preservation standards;
• Artificial intelligence and machine learning;
• Genomic, biometric, imaging, and digital-health data;
• Research-integrity and reproducibility practices;
• Institutional, funder, and regulatory requirements.

Authors are responsible for consulting the current version of this policy and for ensuring that the data-sharing arrangements described in their manuscripts remain accurate and operational.