How Longevity Progress works

In short

• Research comes from public databases every morning: PubMed, ClinicalTrials.gov, bioRxiv, medRxiv and the FDA.
• News and expert views come from a fixed list of trusted outlets, journals and science media centres, quoted word for word.
• “The basics” on each treatment come only from reference sites such as the NIH, MedlinePlus and FDA drug labels.
• Tests, clinics and products are described from the companies' own pages, with outside reviews and FDA notices beside them.
• Books come from book databases and publishers, with reviews from major publications.
• Everything we show is checked against its source and linked to it. Anything that fails a check is left out, not guessed at.
• An AI model helps find and rewrite material in plain English, but it can only use text we fetched ourselves, and it never decides a research stage.
• Nobody can pay to change a research stage, a count, a quote or the order of a list.

Research

Every morning an automated job reads these public sources and adds what is new:

• PubMed, the US National Library of Medicine index of published research.
• ClinicalTrials.gov, the US registry of clinical trials.
• bioRxiv and medRxiv, where researchers post preprints before peer review. We label every preprint as one.
• The FDA press announcements feed.

An item is included when its title or abstract is about aging or longevity, using a published keyword list. Every item links to its original source. Editorials and comments that journals publish about a study are linked too, with their authors named.

News and expert views

We look for what named experts have said about each treatment and about the week's main studies, but only in a fixed list of trusted sources: news organizations, medical journals, science media centres (which collect expert reactions for journalists), hospitals, universities and US government sites.

A quote is shown only if all of these are true: we fetched the page ourselves and the quote appears on it word for word; the person is named, with their role and where they said it; and the quote is about that treatment or study. We quote briefly, never change the wording, and always link to the source. Quotes that fail are dropped and logged.

How “The basics” is written

The basics on each treatment page (what it is, side effects, interactions, approval status, and for clinic treatments what to expect and what it costs) come only from reference sources, and, where those say little, from published reviews of studies in PubMed.

An AI model rewrites them in plain English. Every sentence must match a passage that appears word for word on the source page, and a second check confirms the sentence says nothing more than that passage. Sentences that fail are left out; prices appear only when a source states them. If a source lists side effects or interactions that didn't pass the checks, the whole section is held back rather than shown incomplete. If too little passes, the basics aren't shown at all until a later attempt succeeds. Each section links to its sources.

Tests

Each test's description, price and what you receive come from the company's own pages. We check the company link works and leads to that test, and price changes are checked by a person before they appear. “Research on this test” lists PubMed papers that name the test or its company.

Outside reviews are kept apart from that research. A review is shown only if it is an editorial piece by a named writer on a trusted outlet, is about that test specifically, isn't an affiliate “best of” list, coupon page, press release or paid placement, and the passage we show appears on the page word for word. We don't copy anyone's ratings.

Clinics

Clinic listings come from each clinic's own website: its description, each service in its own words with any price it lists, its locations, and whether it offers telehealth. Everything we show appears word for word on the clinic's pages. Beside a service we link the relevant US regulator page where one exists, such as an FDA consumer warning or approval, shown only when that page is about that exact treatment. A listing is not a recommendation, and nothing on this site says a clinic is safe or unsafe. Registered trials linked to a clinic come from ClinicalTrials.gov.

Products

Products are read from each maker's own product page. Names, ingredients, forms and doses must appear on that page word for word; a product whose page can't be read, or that raises a question, waits for a person to check it. Products don't get a research stage of their own: each shows the stage of its ingredients. We don't rank products, and makers are listed in alphabetical order.

Books

Book details come from Open Library and Google Books, or from the publisher's own page when those have no description. Reviews come from major publications, and every passage we show is checked word for word against the review. Books by anyone connected to Longevity Progress are marked as such, and where a book's details come from elsewhere (for example a store listing) the page says where.

How research stages work

A research stage shows how far testing in people has gone. It does not say whether something works or whether you should use it.

Stages are set automatically every morning from PubMed, ClinicalTrials.gov and FDA records, using these rules in order. Each treatment page says which rule placed it.

• Stage 4 · Widely recommended. Approved by the FDA for that use, or recommended in a major health guideline. We link the approval or guideline.
• Stage 3 · Large studies in people. At least one large (phase 3) trial in people, or at least five large randomized trials, has been published.
• Stage 2 · Small studies in people. There is at least one trial in people, published or registered, but no large ones yet.
• Stage 1 · Tested in animals only. Only studies in animals or cells have been found so far.

“Large randomized trials” means PubMed records tagged as randomized controlled trials that are also multicenter or phase 3, and only trials whose title is about the treatment itself and about aging or a use its goal pages cover. When a search counts a large trial that, on reading, isn't a trial of the treatment for that use, we cap the stage and the page names the paper we checked. An FDA “clearance” of a device is not an approval and doesn't count for stage 4. Products never get a stage of their own.

Under every stage we show the counts behind it, refreshed each morning:

• Randomized trials: PubMed records tagged as randomized controlled trials in people; large ones are also multicenter or phase 3.
• Human trials: PubMed records for the treatment tagged as a clinical trial and as research in people.
• Observational: PubMed records tagged as observational, cohort, case-control or cross-sectional research in people.
• Lab and animal: PubMed records indexed as animal-only or in-vitro research.
• Registered trials: ClinicalTrials.gov interventional listings that match the treatment.

Each treatment page links to the exact searches, so anyone can check the numbers. Counts are search results: they include some records that only mention a treatment, and very new papers may not be tagged yet.

What appears automatically, and what waits for a person

Research items, expert quotes, outside reviews, explainers, clinic details and book details publish automatically once they pass every check above. Whatever fails is dropped and logged, never shown. New or changed prices, and products with an open question, wait for a person to check them. Readers can report a problem on any item, and every report is read.

The role of AI

We use Claude, an AI model made by Anthropic, for four jobs: finding candidate pages to read, pointing to the passages that matter on pages we fetched ourselves, writing plain-English headlines and summaries of studies and trials, and writing “The basics” from reference pages. AI summaries are labeled, restate the source without judging it, and link to the original. The AI never sets a research stage or a count, and any passage it points to must appear on the page word for word before we use it.

Editorial standards

We do not write expert opinions. When we show what an expert or reviewer said, we name the person, give their role and where they said it, and link to the source. Longevity Progress does not employ medical reviewers and does not give medical advice. Talk to your doctor before you change any treatment.

Advertising and affiliate links

Some links to tests and products may earn us a commission. Those links are marked. Sponsored content is always labelled and kept apart from research and news. Companies cannot pay to change a research stage, a count, a quote, our coverage or the order in which we list things. Advertising and sponsorship.

Corrections

If something is wrong, tell us through the contact form or email support@longevityverdict.com with the page and what should change. We correct errors and note the change on the page.