Instructions for Authors

Word limits exclude the abstract, references, tables and figure legends. They support clarity and an efficient review; where the subject genuinely requires more, write to the editorial office before submitting rather than exceeding the limit unannounced.

Keywords: three to eight terms, chosen where possible from a controlled vocabulary such as MeSH, and not repeating words already in the title.

Title page — a separate file

Title; all authors with affiliations and ORCID iDs; the corresponding author's contact details; and the declarations in section 07. Keeping these on a separate file is what makes a double-blind request workable.

Title

A title states what was studied, in what population or setting, and by what approach. Abbreviations are avoided, and a claim of clinical benefit does not belong in a title unless the study design supports it.

Abstract

A structured abstract uses Background, Methods, Results and Conclusions. Report the actual numbers — cohort size, principal estimate with its confidence interval, primary metric — rather than describing that results were obtained. The conclusion must not exceed what the results establish.

Use the structure below for empirical Original Research. Adapt section headings to methodological or formal work; reviews, technical reports and perspectives should use a structure suited to their purpose. Report data, cohorts, estimates and software only where relevant to the study.

Main text

Preparing for double-blind review

Single-blind by default; double-blind review is available on request. A request is made at submission, and it changes what the reviewer manuscript may contain.

For double-blind review, remove author names, affiliations, acknowledgments, funding details, institution names, and other identifying information from the reviewer manuscript. Provide identifying information separately in the title-page file or submission form.

Manuscript file

An editable word-processor format (.docx or .rtf) or LaTeX with a compiled PDF. Continuous line numbers and page numbers throughout.

Language

American English, in a consistent style. Define every abbreviation at first use and use it consistently thereafter.

Figure files

One file per figure, supplied separately. TIFF, EPS, PDF or high-quality PNG. Photographs and screenshots at 300 dpi or better; line art and diagrams at 600 dpi or as vector.

Figure legends

In the manuscript, not in the image. Each legend explains the figure without reference to the text and defines every symbol and abbreviation in it.

Screenshots

Interface captures must contain no real patient data. Use synthetic or fully de-identified content, and say in the legend which it is.

Tables

Editable text, one table per numbered block, with a caption above and footnotes below. Never supplied as an image.

Supplementary files

Cited in the text in order, each with a short title. Protocols, questionnaires, model cards, code listings and extended results belong here rather than in the main text.

Permissions

Written permission compatible with CC BY 4.0 for any reused figure, table, instrument or extended quotation, obtained before submission and acknowledged in the caption.

Number references in the order they first appear, cite them in the text as superscript or bracketed numerals, and list them in that order. Use the journal's own abbreviation as given by the source. Include a DOI wherever one exists. Every reference must be verifiable, and references generated by an AI tool must be checked against the source before submission.

Where a reporting guideline applies to the study design, complete its checklist and submit it with the manuscript, giving the manuscript page or section for each item. Use only the guideline that fits the design — a checklist attached to a study it was not written for tells a reviewer nothing.

These apply where they are scientifically appropriate to the work. A requirement that does not fit the design is addressed by saying why, not by ignoring it.

Decision-science, formal and conceptual studies

State the healthcare decision or information problem and the intended contribution. Decision-science work may examine preferences, uncertainty, shared decisions or decision quality without new software. Report alternatives, outcomes, assumptions, parameter sources and uncertainty or sensitivity analyses where applicable. Formal and conceptual work should present a clear argument or derivation and explain its evaluation and limitations. A clinical cohort, deployment or patient-outcome study is not required when it is not needed to support the claim.

Machine-learning and AI studies

1. The dataset: source system, period, population, size, and the label or reference standard, with how it was derived.
2. The split: how training, validation and test sets were separated, and at what unit — patient, encounter or institution — so that leakage can be ruled out.
3. External validation on data from a different site, period or population where the claim extends beyond the development setting.
4. Comparators: the current practice, clinical score or simpler model the result is measured against. A model reported without a baseline cannot be interpreted.
5. Metrics appropriate to the task and the class balance, with uncertainty, and calibration where the output is used as a probability.
6. Performance across demographic and clinical subgroups, and what the differences mean for the intended use.
7. Interpretability appropriate to the intended user, where the output is intended to inform a decision.
8. Reproducibility: preprocessing, hyperparameters, training procedure, random seeds, and the code and model version.
9. Limitations, stated as limitations of the claim rather than of the field.

Clinical decision-support studies

1. The decision the work supports, and the point in care at which it is made.
2. The intended user — physician, nurse, pharmacist, patient — and what they are expected to do with the output.
3. The comparator: what the decision rests on today.
4. How the recommendation is delivered inside the workflow, and what happens when it is overridden.
5. The evaluation design, and whether it measures model performance, user behavior, or care outcomes. State which.
6. Alerting burden and failure modes, including the consequences of a false positive and a false negative in that setting.

Health-data and interoperability studies

1. Data origin and governance: the system, the custodian, and the approvals under which the data were used.
2. Access: whether another researcher can obtain the data, and how.
3. De-identification method and the residual re-identification risk.
4. Data quality: completeness, missingness and how it was handled, and any known recording artifacts of the source system.
5. Terminologies, code systems and versions used, and how mappings between them were derived and validated.

Informatics system and implementation studies

1. What the system does, and the problem it was built to address.
2. Architecture and standards used, at the level of detail another team would need to build on it.
3. The implementation setting: organization type, scale, existing systems and the period of deployment.
4. The evaluation design, including who was studied and against what comparison.
5. Usability and workflow effects, measured rather than asserted.
6. What did not work, and what would need to be true elsewhere for the result to transfer.

These appear on the title page and, for the published article, after the discussion.

Ethics approval

Committee or IRB and approval reference for any study involving patient data, or the waiver and the body that granted it.

Consent

How consent was obtained or waived, and written consent for publication where any individual could be identifiable.

Trial registration

Interventional clinical trials must be registered in an accepted public trial registry at or before the first participant provides consent for enrollment. The registry name, registration number, and registration date must be reported in the manuscript.

Data availability

Every empirical research manuscript must include a data-availability statement. It names where the data are, on what terms, and the process for obtaining them.

Code availability

Repository, identifier and version for custom algorithms and models, or why they cannot be released.

Author contributions

What each named author did, in enough detail to support the authorship criteria.

Funding

All sources with grant numbers, and the funder's role in design, analysis, interpretation or the decision to publish.

Competing interests

Financial and non-financial interests for every author, including relationships with health IT, EHR, device and AI companies whose products are discussed.

AI use

Any generative AI tool used in preparing the manuscript or conducting the analysis, named, with what it was used for. Authors remain responsible for all content.

Acknowledgments

Contributions that do not meet the authorship criteria, named with permission.

1. The work sits within the journal's scope, and the article type matches what the manuscript is.
2. Word count, abstract length, figure and table count and reference count are within the limits for that type.
3. Title page is a separate file and carries every declaration in section 07.
4. Where double-blind review is requested, the reviewer manuscript carries no identifying information.
5. Line numbers and page numbers are on, and abbreviations are defined at first use.
6. Figures are separate files at the required resolution, with legends in the manuscript and no real patient data in any screenshot.
7. Tables are editable text with captions above.
8. References are numbered in order of appearance, formatted as in section 04, and every one has been checked against its source.
9. The applicable reporting checklist is completed with manuscript locations.
10. The requirements in section 06 that apply to this study design are addressed in the manuscript.
11. Data and code statements name a location, a version and an access process.
12. Permissions are held for all reused material.
13. One submission route has been chosen for this manuscript.

Submission routes