Solutions for getting published

Substantive, AI-assisted help at every step — peer review, editing, proofreading, and submission.

• AI peer review for PhD students

  Build scholarly publishing confidence. Patient, referee-grade draft critique to evaluate methods, argument flow, and claim-evidence alignment.

• AI peer review for academic manuscripts

  Get instant, referee-grade critique of your paper with our advanced AI peer review engine.

• AI peer review for assistant professors

  Accelerate your tenure-track publishing pipeline. Rigorous, referee-grade draft critique to strengthen methods, logic, and evidence before submission.

• AI peer review for clinical trial manuscripts

  Five custom reviewer agents for clinical trial manuscripts: response criteria, dose escalation safety, registration, synergy claims and translational dose.

• AI peer review for epidemiology and population science

  Five prebuilt custom reviewer agents for population research: health equity reporting, competing risks, registry data limits, causal language and qualitative rigour.

• AI peer review for genomics and epigenetics papers

  Five prebuilt custom reviewer agents for omics manuscripts: multiple testing, data availability, batch effects, metabolomics reporting, and variant nomenclature.

• AI peer review for independent researchers

  Dignified, peer-grade critique for scholars in industry and private practice. Stress-test your manuscript without institutional access.

• AI peer review for multilingual & ESL scholars

  Ensure your scientific substance shines through. Substantive, referee-grade evaluation of methods, logic, and data, free from language barriers.

• AI peer review for post-docs

  Exacting manuscript review to secure high-impact publications. Stress-test methods and result framing before submitting to competitive journals.

• AI peer review for preclinical cancer research

  Five prebuilt custom reviewer agents for in vivo and cell biology papers: pseudoreplication, ARRIVE rigour, flow cytometry, image quantification, model relevance.

• AI peer review for research mentors & lab leaders

  Empower your lab with automated, referee-grade manuscript pre-screening. Help trainees self-correct drafts and accelerate lab publishing throughput.

• AI peer review for your paper

  AI peer review that reads like a referee — argument, methods, evidence, and claim consistency checked before your paper reaches the journal.

• AI peer review tool for researchers

  Get instant, referee-grade manuscript analysis using our advanced AI peer review tool.

• AI procurement checklist for research institutions

  Sixteen vendor-neutral questions a research office should ask any AI tool that handles unpublished manuscripts, with what a good answer looks like.

• AI proofreader built for academic drafts

  Move beyond lightweight grammar checking with an AI proofreader built for technical prose.

• AI reviewer for RECIST and response criteria

  A prebuilt custom agent that checks RECIST and iRECIST usage, confirmed responses, censoring rules, and whether the waterfall plot matches the reported response rate.

• AI reviewer for academic papers

  Stress-test your experimental methodology and logic using our professional AI reviewer.

• AI reviewer for batch effects and replicate structure

  A prebuilt custom agent that checks whether batch is confounded with condition, whether batch was modelled, and whether replicates are biological or technical.

• AI reviewer for causal language in observational studies

  A prebuilt custom agent that finds causal verbs applied to associational findings and checks whether the adjustment set is justified rather than assembled.

• AI reviewer for competing risks and time-related bias

  A prebuilt custom agent that checks Kaplan-Meier versus Fine-Gray, immortal time in registry cohorts, and lead-time and length-time bias in screening studies.

• AI reviewer for data availability and accessions

  A prebuilt custom agent that checks GEO, SRA and dbGaP accessions, genome build, tool versions and code availability before you submit an omics manuscript.

• AI reviewer for dose escalation and safety reporting

  A prebuilt custom agent that checks the DLT definition and window, CTCAE version, safety denominators, and whether the recommended phase 2 dose is justified.

• AI reviewer for drug synergy claims

  A prebuilt custom agent that checks whether a synergy claim names a reference model — Bliss, Loewe, HSA or Chou-Talalay — or rests on a bar chart.

• AI reviewer for ethics and data sharing statements

  A prebuilt custom agent that checks ethical approval, consent scope for data sharing, de-identification claims and data management plan compliance.

• AI reviewer for flow cytometry reporting

  A prebuilt custom agent that checks gating strategy, FMO and viability controls, compensation, and the shifting denominators behind reported cell percentages.

• AI reviewer for health equity reporting

  A prebuilt custom agent that checks how race, ethnicity and socioeconomic position are measured, justified and interpreted in health disparities research.

• AI reviewer for image quantification integrity

  A prebuilt custom agent that checks sampling, blinded scoring, thresholding method and scale bars behind IHC, immunofluorescence and microscopy quantification.

• AI reviewer for in vivo experimental rigour

  A prebuilt custom agent that checks randomisation, blinding, sample size justification, exclusions, animal sex and tumour volume reporting in animal studies.

• AI reviewer for metabolomics reporting standards

  A prebuilt custom agent that checks normalisation, internal standards, isotope natural-abundance correction and metabolite identification confidence levels.

• AI reviewer for mouse model relevance

  A prebuilt custom agent that checks whether immune and microenvironment claims are supportable in the model used, from nude xenograft to syngeneic to humanised.

• AI reviewer for multiple testing in omics papers

  A prebuilt custom agent that checks FDR correction, enrichment background universes, and ranking metrics in RNA-seq, ChIP-seq and screen manuscripts.

• AI reviewer for overclaiming in research papers

  A prebuilt custom agent that measures the distance between what a study showed and what its abstract and discussion say, and proposes replacement wording.

• AI reviewer for pseudoreplication and unit of analysis

  A prebuilt custom agent that finds n counted as cells, tumours or wells when the independent unit is the animal or the experiment, and nested data run through a t-test.

• AI reviewer for qualitative and mixed-methods rigour

  A prebuilt custom agent that checks sampling, saturation, coding reliability, reflexivity and member checking in interview and focus group manuscripts.

• AI reviewer for registry and claims data limitations

  A prebuilt custom agent that checks whether SEER, NCDB and claims analyses acknowledge coding validity, missing treatment detail, and selection into the data source.

• AI reviewer for translational dose relevance

  A prebuilt custom agent that checks whether in vitro concentrations are clinically achievable and whether animal-to-human dose conversion has a stated basis.

• AI reviewer for trial registration and endpoints

  A prebuilt custom agent that checks the registration identifier, whether the reported primary endpoint matches the registered one, and outcome switching.

• AI reviewer for variant and gene nomenclature

  A prebuilt custom agent that checks HGVS variant syntax, versioned transcript references, and current HGNC gene symbols across a genomics manuscript.

• AI reviewer that reads like a grant study section

  A prebuilt custom agent that reads a manuscript the way an NIH reviewer scoring significance, innovation and approach would — a different job from peer review.

• AI that never trains on your research

  How PerfectPaper keeps your manuscript out of model training: its own code trains on nothing, providers are bound by contract terms, and every processor is named.

• AI-assisted peer reviewer for draft critique

  Exacting, constructive peer reviewer feedback on your draft in minutes. Built for academic science.

• Academic editing for researchers

  Academic editing that goes beyond grammar — AI-driven substantive feedback on argument, structure, clarity, and journal-style fit for your manuscript.

• Academic manuscript editing services

  Dignified, peer-grade manuscript editing services for active scientific authors.

• Academic peer review software for authors & groups

  Collaborative, automated academic peer review software designed specifically for research groups.

• Academic proofreading

  Academic proofreading that goes beyond grammar — PerfectPaper gives researchers substantive, peer-reviewer-grade feedback before submission.

• Academic writing feedback for researchers

  Academic writing help that reads like a reviewer, not a grammar engine — PerfectPaper checks structure, argument, evidence, and clarity before you submit.

• Best proofreading services for researchers

  Compare the best proofreading services for academic manuscripts and doctoral drafts.

• Can peer reviewers use AI on manuscripts?

  Most publishers prohibit it, and the reason is custody rather than quality: uploading a manuscript under review is a distribution the author never consented to.

• Custom AI reviewer agents for research papers

  Add up to three custom reviewer agents to a PerfectPaper review. Browse prebuilt briefs by field, or write your own for the conventions your subfield cares about.

• Desk rejected without review

  The fastest rejection and usually the most fixable. The five causes, how to tell which one you got from a form letter, and what to change before resubmitting.

• Desk screening at scale

  The fastest way to protect reviewer capacity is to send fewer papers to review. What can be checked before an invitation goes out, and what cannot.

• Dissertation editing services for PhD candidates

  Professional dissertation editing services to polish methods, chapters, and formatting before defence.

• Dissertation proofreading & editing

  AI-assisted dissertation proofreading and editing for PhD students — substantive feedback on your full manuscript, not just grammar.

• Dissertation proofreading services

  High-accuracy dissertation proofreading services to verify terminology and references.

• Do journal AI policies apply to authors or referees?

  Journal AI policies split by role: referees are prohibited from uploading a manuscript to an AI tool; authors may use AI on their own work and disclose it.

• False discovery rate or family-wise error rate?

  Family-wise error rate bounds the chance of any false positive. False discovery rate bounds the expected share of false ones among your findings. The claim decides.

• Get your paper ready for journal submission

  Catch the issues that cause desk rejects before you submit. PerfectPaper gives your manuscript a referee-grade review for journal submission.

• How do you write a limitations section?

  A limitations section names each specific threat to validity, states the direction and magnitude of the bias it would cause, and tempers the conclusion accordingly.

• How do you write a research paper abstract?

  A research paper abstract is written last, from the finished manuscript, in six moves. Give the primary result as an effect size with precision, not an adjective.

• How to know if an article is peer reviewed

  Learn how to know if an article is peer reviewed using indexing databases and journal policies.

• How to write a cover letter to a journal editor

  A cover letter states what the paper claims, why this journal, and what must be disclosed. The required statements, a working structure, and what to leave out.

• How to write a custom AI reviewer agent

  A structure that works for AI peer review prompts: scope, ordered checks, what to report, and the negative instructions that stop an agent inventing findings.

• How to write a response to reviewers

  The response letter decides more revisions than the experiments do. Structure, tone, how to disagree, and the four mistakes that earn authors a second round.

• Is it safe to upload an unpublished paper to ChatGPT?

  Check three things before uploading an unpublished manuscript to an AI tool: your plan's training default, its retention window, and your institution's policy.

• Journal selection: is your paper right for the journal?

  Choosing a journal is only half the decision — PerfectPaper checks whether your manuscript fits the journal's scope and standards before you submit.

• Manuscript editing for researchers

  Manuscript editing for academic researchers — AI feedback on argument, methods, and structure to fix what causes journal rejections.

• Mediator or confounder?

  A confounder causes the exposure; a mediator is caused by it. Adjusting for a confounder removes bias, adjusting for a mediator removes the effect you are measuring.

• My effect disappeared when I added a covariate

  Adjustment removing an effect can mean confounding was controlled — or that you adjusted for a mediator or a collider and deleted the pathway you were studying.

• My knockdown and knockout phenotypes disagree

  Usually genetic compensation, off-target effects, or the difference between acute and chronic loss — and which one it is changes what you can claim.

• My p-value gets smaller every time I measure more cells

  If measuring more cells from the same animals keeps improving significance, the test is counting measurements as independent observations and the p-value is not real.

• My replicates disagree

  Whether disagreement is a problem depends on what kind of replicate disagreed. Technical spread is a measurement issue; biological spread is usually the finding.

• My survival curves cross

  Crossing curves mean the hazard ratio you reported is an average of effects that reversed direction, and the log-rank test is at its least powerful exactly there.

• Rejected after review — what to do next

  Triage the reviews before choosing a journal. Which criticisms will recur anywhere, which were venue-specific, and how to tell a fatal objection from a fixable one.

• Research methods concepts, explained plainly

  Short, precise explanations of the bias, design and inference terms that decide whether a paper survives review — each with how to detect it in your own work.

• Responding to novelty and scope objections

  Objections about significance are judgements, not defects. Which are answerable by reframing, which by new work, and which mean the venue is wrong.

• Responding to reviewer comments on methods and design

  What reviewers mean when they ask for more controls, an orthogonal method, or mechanism — and which of those requests can be answered without new experiments.

• Responding to statistical reviewer comments

  What reviewers mean by the twelve most common statistical objections, what they want to see, and how to answer when you can fix it and when you cannot.

• Response to reviewers sample and template

  Download our proven response to reviewers sample and template to handle referee comments with ease.

• Reviewer asked for a post hoc power calculation

  Power computed from your observed effect just restates your p-value. Why it answers nothing, what the reviewer actually wants, and what to send instead.

• Reviewer questions my choice of statistical test

  Normality, pairing and unequal variance are the three grounds behind most of these comments. What each requires, and when switching tests is the wrong fix.

• Reviewer questions whether the experiment was replicated

  Usually a reporting gap rather than a missing experiment. What to state in the legend, and why 'representative of three independent experiments' causes the comment.

• Reviewer says my n is not independent

  Pseudoreplication is the one statistical objection that changes p-values by orders of magnitude, and the remedy is almost always reanalysis rather than new experiments.

• Reviewer says my sample size is too small

  The comment usually means one of three different things. Work out which before you respond, because the answer to each is different and only one needs new data.

• Reviewer says the effect is not clinically meaningful

  Statistical significance is not magnitude. This objection is about whether the size of your effect matters, and more data makes it worse rather than better.

• Reviewer says the mechanism is not shown

  Often answerable by rewriting rather than by experiments. The comment usually means the paper promised a mechanism it did not deliver, not that one is required.

• Reviewer says the model system is not appropriate

  The objection that most often ends in a narrower claim rather than a new experiment. What the model can support is a question about the sentence, not the science.

• Reviewer says the paper is out of scope

  The fastest objection to act on, because it is rarely worth arguing. How to tell a genuine scope mismatch from a reviewer using scope for another reason.

• Reviewer says the results are correlative, not causal

  A standing objection in molecular biology. What counts as causal evidence, and why rewriting the claim is often the correct answer rather than a retreat.

• Reviewer says the work is incremental

  A judgement about framing as often as about content. What incremental usually means, and the three ways a paper gets read that way when it should not be.

• Reviewer says this is already known

  Check what the cited prior work actually showed before conceding. The claim is frequently narrower than the reviewer remembers, and that difference is your paper.

• Reviewer wants an orthogonal method

  Orthogonal validation is a demand about the measurement, not the analysis. Which second method counts depends on which failure mode the reviewer has in mind.

• Reviewer wants more controls

  Which control the reviewer means is usually inferable from the rest of the review, and one whole class of this objection is answered with text rather than experiments.

• Reviewer wants multiple comparison correction

  When correction is genuinely required, when it is arguable, and how to answer a reviewer who asks for it on analyses where it would be the wrong choice.

• Scholarly proofreading and editing services

  Substantive copy editing and proofreading and editing services to prepare your paper for publication.

• Scholarly proofreading services

  Calibrated proofreading services for academic researchers. Verify terminology and references.

• Scientific manuscript editing services

  Exacting scientific manuscript editing services to prevent journal desk rejections.

• Secure AI review for unpublished research

  How PerfectPaper handles an unpublished manuscript: never used to train a model, processed under named terms, stored in the EU, every processor named.

• Something looks wrong with my results

  Symptom-first diagnosis for analysis surprises: an effect that vanishes on adjustment, curves that cross, replicates that disagree, a p-value that keeps falling.

• Statistical review capacity

  Most journals cannot get a statistician for most papers, so statistical problems are found by subject reviewers who were not asked to look — or after publication.

• The peer review capacity problem

  Submissions grow, the invited reviewer pool does not, and decline rates rise. The levers editors actually control, and where computational checking legitimately fits.

• Thesis editing services for master's candidates

  Ensure cross-chapter logical consistency and formatting with our thesis editing services.

• Thesis proofreading services

  Exacting thesis proofreading services to eliminate notation and styling inconsistencies.

• Thesis vs dissertation: what is the difference

  Understand the difference between a thesis vs dissertation across different graduate schools and countries.

• What are the ARRIVE guidelines?

  The ARRIVE guidelines are a 21-item checklist for reporting animal research, published by the NC3Rs in 2010 and revised as ARRIVE 2.0 in 2020 with an Essential 10.

• What is Simpson's paradox?

  Simpson's paradox is a reversal: an association in every subgroup flips or vanishes when subgroups are pooled. Both numbers are correct; one answers the question.

• What is a Kaplan-Meier curve?

  A Kaplan-Meier curve is the product-limit estimate of survival: it steps down at each event and counts censored patients for as long as they were actually followed.

• What is a batch effect?

  A batch effect is systematic technical variation between samples processed together. It biases results when batch correlates with the condition studied.

• What is a composite endpoint?

  A composite endpoint counts a participant as having an event when any one of several specified outcomes occurs, such as death, myocardial infarction or stroke.

• What is a confidence interval?

  A confidence interval is a range produced by a method whose coverage holds over repeated studies. The 95% describes that method, not the single interval reported.

• What is a directed acyclic graph in epidemiology?

  A directed acyclic graph is a diagram of causal assumptions: nodes are variables, arrows are direct causes, no path loops back. It decides the adjustment set.

• What is a funnel plot?

  A funnel plot charts each study's effect estimate against its precision in a meta-analysis. Asymmetry means small-study effects, of which publication bias is one cause.

• What is a hazard ratio?

  A hazard ratio compares the instantaneous event rate in two groups among those still at risk. It is a rate ratio, not a risk ratio, and says nothing about timing.

• What is a minimal clinically important difference?

  A minimal clinically important difference is the smallest score change patients experience as beneficial. It belongs to a population and context, not the instrument.

• What is a non-inferiority trial?

  A non-inferiority trial tests whether a new treatment is worse than an active control by more than a prespecified margin. The margin, not the p-value, carries the claim.

• What is a surrogate endpoint?

  A surrogate endpoint stands in for an outcome patients feel. It is valid only when treatment effects on the surrogate predict treatment effects on the real outcome.

• What is an effect size?

  An effect size states how large a difference, ratio or association is, on a scale a reader can judge. A p-value cannot supply that number, and never has.

• What is collider bias?

  Collider bias is created by adjustment rather than removed by it. Conditioning on a variable caused by both exposure and outcome induces an association that is not real.

• What is confounding?

  Confounding occurs when a common cause of both exposure and outcome distorts their association. It exists in the data before analysis, unlike collider bias.

• What is heterogeneity in meta-analysis?

  Heterogeneity in meta-analysis is variation in the true effects across studies, beyond sampling error. I² gives its share of total variation; τ² gives its size.

• What is immortal time bias?

  Immortal time bias arises when exposure is defined by an event patients had to survive to reach, guaranteeing the exposed group appears to live longer.

• What is intention-to-treat analysis?

  Intention-to-treat analysis compares participants in the groups they were randomised to, whatever treatment they received: the effect of offering treatment.

• What is lead-time bias?

  Earlier detection lengthens the interval between diagnosis and death without delaying death, making screened patients appear to survive longer when nothing changed.

• What is length-time bias?

  Screening preferentially detects slow-growing disease, which spends longer in a detectable preclinical state. Screen-detected cases have better prognosis by selection.

• What is multiple testing correction?

  Multiple testing correction adjusts thresholds or p-values when many hypotheses are tested. FWER methods control any false positive; FDR methods control the proportion.

• What is overdiagnosis?

  Overdiagnosis is the correct detection of disease that would never have caused symptoms or death. It is defined at population level, not in one patient.

• What is overfitting?

  Overfitting is a model learning noise specific to its own sample, so apparent performance overstates performance in new data. The gap is called optimism.

• What is p-hacking?

  P-hacking is trying analytic options until one crosses p < 0.05 and reporting only that one. Four common flexibilities together give a 60.7% false-positive rate.

• What is propensity score matching?

  Propensity score matching pairs treated and untreated subjects with similar probability of treatment. It balances measured covariates only, never unmeasured ones.

• What is pseudoreplication?

  Pseudoreplication counts measurements from one experimental unit as independent. It inflates significance by orders of magnitude and is detectable without statistics.

• What is publication bias?

  Publication bias arises when whether a study is published depends on what it found. Positive results appear more often and sooner, so the literature overstates effects.

• What is regression to the mean?

  Regression to the mean is the tendency of an extreme measurement to be followed by a less extreme one, because selection captures transient error alongside real signal.

• What is selection bias?

  Selection bias comes from who entered a study and who stayed in it, not from a common cause inside it. Adjustment does not remove it, and a larger sample does not.

• What is statistical power?

  Statistical power is the probability a design detects an effect of a stated size. Power is fixed before data exist, which is why power computed afterwards is circular.

• What is the CONSORT checklist?

  The CONSORT checklist is the reporting standard for randomised trials. CONSORT 2025 has 30 items and a participant flow diagram, and governs the report, not the trial.

• What is the COREQ checklist?

  COREQ is a 32-item reporting checklist for interview and focus group studies, in three domains: research team and reflexivity, study design, analysis and findings.

• What is the PRISMA checklist?

  The PRISMA checklist is a 27-item reporting guideline for systematic reviews and meta-analyses. It specifies what a review must report, not how well it was conducted.

• What is the STROBE checklist?

  STROBE is a 22-item reporting guideline for observational studies: 18 items common to cohort, case-control and cross-sectional designs, four design-specific.

• What is the Table 2 fallacy?

  Presenting every coefficient in one adjusted model as a causal effect. A single adjustment set can be correct for the exposure or for a covariate, but not for both.

• When to appeal a rejection

  Appeals rarely succeed, and there is a reliable test for the cases where they do: a factual error a third party can verify without re-litigating judgement.

• Where your manuscript is processed

  The route an uploaded manuscript takes through PerfectPaper: Frankfurt servers, EU storage, named model providers, and the parts that do not run in Europe.

• Why it is so hard to find peer reviewers

  Decline rates rise because invitations concentrate on the same people. What editors can change: invitation targeting, reviewer load, and what reaches review at all.

• Why papers get rejected, and what to do next

  Most rejections come from a small number of causes, and which one you received determines whether to revise, resubmit elsewhere, appeal, or restructure.