How MedBrief Works

From a study published in a medical journal to a plain-English summary in your inbox — here's exactly what happens.

Step 1 — Finding new studies

Every day, MedBrief automatically checks two databases for newly published research:

  • PubMedThe world's largest database of peer-reviewed biomedical literature, maintained by the US National Library of Medicine. It indexes tens of thousands of journals.
  • ClinicalTrials.govThe official US registry of clinical trials, maintained by the National Institutes of Health. It covers ongoing and completed trials worldwide.

We search using carefully chosen medical terms related to each condition we track, so we catch studies that use different terminology for the same condition.

Step 2 — Selecting relevant studies

Not every study that mentions a condition is worth reading about. We apply a two-pass filter before a study gets summarised.

Basic eligibility criteria

A study must meet all of the following to be included:

  • Published in English
  • Peer-reviewed (published in an indexed journal or registered as a trial)
  • Empirical — meaning it reports actual data, measurements, or outcomes
  • Has an abstract we can read and summarise

We automatically exclude the following:

  • Editorials and opinion pieces
  • Letters to the editor
  • Studies in animal models only (no human data)
  • Studies with no discernible link to the condition

Relevance scoring

After passing the basic filter, each study is given a relevance score that determines how prominently it appears on MedBrief. The score is a weighted combination of four factors:

Study type

35%

Randomised controlled trials and meta-analyses score highest. Cohort studies and clinical trials score moderately. Observational and case reports score lower.

Sample size

25%

Studies with more than 100 participants score highest. 30–100 scores moderately. Under 30 scores lower, and we flag it clearly.

Recency

25%

Studies published in the last 6 months score highest. The score gradually decreases for older studies.

User ratings

15%

Thumbs up and thumbs down from our community. Only applied once a study has at least 5 ratings.

The “Most Relevant” sort on the browse page uses this score. You can also sort by Most Recent, Highest Rated, or Oldest First.

Step 3 — Quality badges

Each study displays quality badges so you can assess its strength at a glance, without reading anything.

Sample size

n > 100n = 30–100n < 30

Larger studies are generally more reliable — their results are less likely to be a fluke.

Study type

Gold StandardHigh QualityModerate

Gold Standard = randomised controlled trials (RCTs) and meta-analyses. These have the strongest evidence because they control for bias.
High Quality = clinical trials and cohort studies.
Moderate = observational and case-control studies.

Methodology

RigorousStandardUnclear

Rigorous indicates double-blind methodology, which further reduces bias in the results.

Industry funding

Industry funded

Studies funded by pharmaceutical companies are flagged. This doesn't make them wrong, but it's something to be aware of when reading the results.

Step 4 — AI summary generation

Once a study passes our filters, it goes through a two-agent AI review process.

1

Agent 1 — Summariser

Reads the study abstract and generates a structured summary with four sections: a quick 2–3 sentence overview, what was studied and why, how the study was conducted, and the key findings as bullet points. It then lists the important limitations.
2

Agent 2 — Reviewer

A second AI agent independently checks the summary. It verifies that numbers match the abstract exactly, that limitations are comprehensive, that the reading level is accessible (high school to college), and that no treatment recommendations have been made. If the summary fails these checks, it sends it back for revision.
3

Up to 2 revisions

If the reviewer rejects the first summary, Agent 1 revises it based on specific feedback. This happens up to twice. If the summary still fails after two revisions, it is flagged for manual review and not published until resolved.

About 91% of summaries are approved without any issue. The process typically costs around $0.06 per study.

Step 5 — Your email digest

You choose whether to receive updates weekly or monthly. At the end of each period, MedBrief sends you a digest with the most relevant new studies from that window — prioritised by relevance score, not just publication date.

Each study in the digest includes the title, quality badges, and the quick summary. You can click through to read the full summary on MedBrief at any time.

You can change your digest frequency or turn off emails entirely from your settings page.

Still have questions?