Import records
Search PubMed directly, or import exports from PubMed, Embase, Scopus or Web of Science. Duplicates are removed by PMID, DOI, then normalised title, and counted for the PRISMA diagram.
Search PubMed
Note: NOT (oral) also drops studies that only mention oral cancer as a comparison group. If you want to be conservative, remove it here and exclude oral-only studies during screening.
NCBI API key (optional)
Without a key, requests are limited to 3 per second; with one, 10 per second. The key is stored only in this browser's localStorage and is never included in project exports.
Import files
Drop files here, or
PubMed / MEDLINE (.nbib, .txt), RIS (.ris), CSV / TSV, PubMed XML
From PubMed: Save → Selection: All results → Format: PubMed gives an .nbib file. CSV needs a header row; recognised columns are title, abstract, authors, year, journal, doi and pmid (PubMed's own CSV export works).
Import history
| Date | Source | Detail | Found | Added | Duplicates |
|---|
Duplicates removed (0)
| Duplicate title | Matched by | Source |
|---|
Full-text review & extraction
Records marked Include or Maybe at screening. Record retrieval, the eligibility decision with a reason, and extract 2×2 data and QUADAS-2 ratings for included studies.
All records
Every unique record with its decisions. Select a title to open it in the screener.
| Title | Year | Source | Screening | Reason | Full text | Full-text reason |
|---|
PRISMA 2020 flow diagram
Computed live from your imports and decisions. Maybe records at screening are treated as sought for retrieval. One report per study is assumed.
Criteria & settings
One item per line. Changes apply immediately. Reasons are stored as text on each record, so editing this list does not rewrite past decisions.
Inclusion criteria (reference)
Include: human case–control or cohort studies of salivary miRNA diagnostic accuracy for non-oral systemic cancers (e.g. pancreatic, lung) reporting sensitivity, specificity or AUC, or data to reconstruct them.
Exclude: oral or head-and-neck-only cancers; no diagnostic accuracy data; non-human; reviews.
Risk of bias: QUADAS-2.