Most students treat PubMed like Google: dump a keyword, open twenty tabs, highlight until the PDF is yellow, and still cannot write the related-work paragraph.

A better first pass is mechanical.

1. Write the question as Population + Intervention/Exposure + Outcome. If you cannot fill those three, you do not have a search yet.

2. Pull 12–20 papers, not 200. Sort once by citations (what the field already believes) and once by year (what just changed).

3. For each abstract, write one sentence that starts with a verb: increased, reduced, failed to, was associated with. If you cannot, the paper does not belong in the review.

4. Cluster those sentences. Contradictions are the review. Agreement is the background.

5. Cite as you go. A bibliography you build at the end is a bibliography you will fake.

Helix is built around that loop — search, extract, cluster, cite — so the writing starts from findings, not from tabs.