Check Claims Before You Publish

Author Review: Alvin C Powers

Evidence boundary (critical): I only received aggregate metrics and a list of paper titles. I was not provided DOIs, URLs, abstracts, methods, or full-text excerpts. Therefore, I cannot verify experimental design quality, statistical rigor, reproducibility, effect sizes, limitations, or whether results are robust. This review focuses on scope signals (what topics the titles suggest) and on epistemic gaps (what would be needed to evaluate rigor properly).

1) What the paper titles suggest (topic scope)

I grouped titles using a transparent, manual keyword heuristic (not a model; not a bibliometric topic model). This is a weak proxy for expertise: titles can be misleading about what was actually tested.

2) What seems “strong” vs “uncertain” from titles

3) Bibliometric signals you provided (interpret cautiously)

You provided: h-index = 4, total citations = 48, paper count = 12. These metrics correlate with impact but are not a direct measure of rigor; they are also sensitive to field size, publication age, and citation practices.

4) Paper-level red flags to check next (what I would audit)

Because I can’t access methods from what was provided, here are the most important rigor audit questions to run on each paper (especially immunology/diabetes studies):

1. Study design: Was the primary endpoint pre-registered or at least pre-specified? Are comparisons properly powered?
2. Controls: Are negative/positive controls adequate (isotype controls for flow cytometry, appropriate tissue/age controls for aging studies)?
3. Blinding & randomization: Are histology/scoring and analysis blinded to group assignment?
4. Batch & donor effects: For human islet/immune data, were donor-level random effects or matched designs used?
5. Multiple comparisons: Immunology screens can inflate false discovery—are adjustments applied?
6. Mechanistic claims: When titles imply mechanism (e.g., “recruits CD8+ T cells” or “mTOR-dependent”), do they provide direct causal evidence (perturbations, pathway blockade/rescue)?
7. Translation limits: For nonobese diabetic mouse results, do authors discuss failure modes in human disease?

Provide DOIs/URLs for each listed paper (or paste abstract + methods + results). Then I can: (i) extract sample sizes and statistical methods, (ii) distinguish correlation vs causation, (iii) evaluate reproducibility signals, (iv) produce evidence-graded claims with inline citations.

Because no DOIs/URLs/full text were supplied, my confidence about experimental rigor and biological validity is low. Confidence is higher only regarding the topic specialization signals implied by the titles.