Artificial Intelligence and Automated Tools

Clinical Medicine and Pharmacology permits responsible investigation and use of artificial intelligence when the methods and limitations are transparent. An AI system cannot assume responsibility for a paper or substitute for the accountable judgment of its authors and editors. The rules below distinguish AI used as a research method from tools used to prepare or assess a manuscript.

Research methods involving AI

Studies using AI for diagnosis, drug discovery, prediction or clinical decision support describe the model, relevant version, training and evaluation data, preprocessing, validation strategy and limitations. Authors report safeguards against data leakage, biased evaluation and misleading claims of clinical readiness. Synthetic patient data must be labelled as synthetic and must not be presented as observations from real patients. Restrictions affecting access to a model or dataset are described in the availability statement.

Assistance in preparing the paper

Authors disclose substantive generative assistance in writing, translation, coding, analysis or illustration, naming the tool and version where available, describing its purpose and identifying the parts affected. They explain the human checks applied. Routine spelling or reference formatting tools do not require a lengthy declaration unless their operation materially changes content. AI cannot be listed as an author, and acknowledging a tool does not transfer responsibility for inaccuracies, plagiarism or privacy breaches.

Unacceptable uses

Tools must not invent references, alter clinical observations, generate purported patient images, conceal contradictory results or create a false account of ethical approval. Illustrations that are genuinely schematic or synthetic must be identified as such and must not imply evidence that was never collected. Authors verify every citation and clinically relevant statement. They must not submit patient data or confidential research material to a service without the authority, safeguards and permissions required for that disclosure.

Editors reviewers and production

Reviewers perform the scientific assessment themselves. They may not upload a confidential submission, supplementary data or draft report to an unapproved external service, nor secretly outsource the review. Any permitted assistance that materially informs a review is disclosed to the handling editor. Editors retain responsibility for decisions. The journal identifies any deployed automated editorial or production functions that materially affect authors and explains the human oversight; no undeclared system decides acceptance.

Assessment and remedy

A detector signal is considered with manuscript evidence, references, data provenance and the author’s explanation. The journal applies separate 30 percent similarity and AI-writing screening limits under the procedure below; a report exceeding a limit requires assessment before the manuscript can progress, but the percentage alone does not establish misconduct. An incomplete disclosure may call for clarification, while fabricated evidence or a serious confidentiality breach may require integrity action. Authors should correct an inaccurate AI declaration as soon as it is discovered. Questions or supporting documents should be sent through the CMP editorial office at cmpeditorialoffice@jandoopress.com, identifying the manuscript number or article DOI where available.

Sources and automated screening

Generative AI output is not cited as an authoritative scholarly source. Where a system itself is the object of research, its output may be documented as study material with suitable methodological detail and labelling. Editors and reviewers must not use generative AI to create their scientific assessments; permitted disclosed language editing must preserve a human written evaluation and confidentiality. Routine journal tools must be tested before use, and a staff member or editor verifies automated flags and reviewer suggestions.

Declared journal screening tools

The journal uses Turnitin for English text for similarity and AI-writing screening. The maximum reported screening percentage is 30 percent for each measure, subject to applicable report requirements and documented human assessment under Originality Plagiarism and Overlapping Publication. An AI result is an estimate of eligible text, not a measurement of authorship or a guaranteed probability of misconduct. Missing, unsupported or inconclusive output is not recorded as a passing score.

The Research Integrity Officer restricts access to reports and manuscript files, verifies the service settings and retains the minimum necessary assessment record. Authors must accurately disclose substantive generative-AI assistance. Editors and reviewers may not use generative AI to create their scholarly assessments or upload confidential patient consent, case narratives and linked clinical records to unapproved services. A justified concern about an already published article follows the journal's investigation, correction and retraction process; confirmed author-attributable serious violations do not entitle an author to an APC refund.

Last updated: 17 September 2026