Reviewer guidelines

Reviewer Guidelines — AIDSAE

Peer review model: Double-blind (authors and reviewers anonymous to each other)

Thank you for agreeing to review for AIDSAE. Your review helps us publish credible, methodologically sound, and practically relevant AI/data science research for health, agriculture, and environmental systems.

1) Before You Accept the Invitation

Please accept the review only if:

A) Expertise and time

  • You have suitable expertise for the topic and methods.
  • You can submit your review within the requested timeline (typically 10–14 days).

B) Conflicts of interest
Declare and decline if you have a conflict, including:

  • recent collaboration with the authors,
  • same institution/department,
  • financial/personal interests,
  • direct competition that could bias judgment.

C) Confidentiality
The manuscript and all review communications are confidential. Do not share, distribute, upload, or use the content for personal advantage.

2) What AIDSAE Expects From Reviews

AIDSAE values specific, evidence-based, and constructive feedback. Please aim to:

  • identify major issues affecting validity, reproducibility, or conclusions,
  • suggest practical revisions,
  • point out missing information required for evaluation,
  • flag integrity risks (e.g., implausible results, invented citations, data leakage).

Avoid vague statements like “methods are weak” without explaining what is missing or how to fix it.

3) How to Structure Your Review (Recommended Format)

A) Summary (3–6 sentences)

  • What the paper attempts to do (data + method + application).
  • What you believe is the main contribution.
  • Your overall view of whether the approach is valid and useful.

B) Major Comments (numbered)

Focus on issues that must be addressed before acceptance. Typical major issues include:

  • unclear dataset provenance or eligibility criteria,
  • weak validation (no test set, improper cross-validation, tuning on test set),
  • no baseline comparator or unfair comparisons,
  • inappropriate metrics or missing uncertainty,
  • suspected data leakage,
  • overclaiming / unsupported conclusions,
  • poor reproducibility (missing pipeline details, no code/data statement).

For each major comment, include:

  • what is wrong/missing,
  • why it matters, and
  • what the authors should do.

C) Minor Comments (numbered)

  • clarity, organization, missing citations, figure/table issues, typos.

D) Recommendation

Choose one:

  • Accept
  • Revision
  • Reject

Add a brief justification (1–3 sentences).

4) AIDSAE Technical Checklist (Use This to Judge Validity)

4.1 Data and provenance

  • Is the dataset source clearly described (where it came from, time period, inclusion/exclusion)?
  • Are access restrictions/permissions explained?
  • If human/sensitive data: are ethics/privacy safeguards described?

4.2 Preprocessing and features

  • Are cleaning steps and missing-data handling explained?
  • Is feature engineering described clearly enough to reproduce?

4.3 Validation and tuning (most common failure point)

Check carefully:

  • Train/validation/test split is clearly defined (or CV design).
  • Hyperparameter tuning is done without leaking test information.
  • Any resampling is appropriate (stratification, grouped splits, temporal/spatial splits where necessary).
  • If repeated measures (patient/farm): ensure group-wise splitting to avoid leakage.

4.4 Baselines and comparisons

  • Do authors compare against at least one meaningful baseline?
  • Are comparisons fair (same data, same splits, same preprocessing)?

4.5 Metrics and uncertainty

  • Are evaluation metrics appropriate and correctly computed?
  • Do they report confidence intervals, multiple runs, or variability where appropriate?
  • Is calibration considered for risk prediction tasks (if relevant)?

4.6 Error analysis and limitations

  • Do authors analyze failure modes or error patterns?
  • Do they discuss limitations, bias, generalisability, and domain shift?

4.7 Reproducibility statement

  • Do they state whether data/code/models are available?
  • If restricted, do they provide sufficient detail for “reproducible in principle”?

5) Responsible AI / Ethics (Flag When Relevant)

Please comment when the manuscript involves:

  • high-stakes health decisions, vulnerable populations, or sensitive geolocation,
  • fairness/bias risks (e.g., performance differs by subgroup),
  • privacy risks (identifiability, re-identification),
  • unvalidated deployment claims (“ready for clinical use” without external validation).

6) Signs That Require Extra Scrutiny (Possible Integrity Issues)

Flag to the editor (in “Confidential comments”) if you notice:

  • results that look too perfect or inconsistent with sample size,
  • identical metrics across folds/runs without variance,
  • suspiciously generic text or mismatched references,
  • missing details that prevent verification,
  • evidence of data leakage (e.g., preprocessing done before splitting),
  • invented citations or references that don’t match claims.

If you suspect misconduct, do not accuse the authors directly in the public comments; instead, explain the concern neutrally and request clarification, and alert the editor confidentially.

7) Confidential Comments to the Editor (Optional but Helpful)

Use this space for:

  • integrity concerns or strong reservations,
  • conflicts or anonymity issues,
  • whether the paper is salvageable with major revisions.

8) Tone and Professional Conduct

  • Be respectful and constructive, even when recommending rejection.
  • Focus on the work, not the authors.
  • Avoid using identifying language (since review is double-blind).

9) Typical Review Timeline

AIDSAE aims for an initial decision within ~3–4 weeks. Reviewers are typically requested to submit reports within 10–14 days. If you need more time, please inform the editorial office as early as possible.

10) Contact

For questions or to request an extension, contact the editorial office: editorialoffice@ecoscribepublishers.com