Is that health claim true?
Paste a caption, a label or a line you heard at the gym. The tool splits it into claims, searches PubMed, keeps only papers from journals with an impact factor of 3 or more, and writes one verdict per claim. It is the same check JD Botezelli PhD runs before he records a reel.
Why not just ask a chatbot?
A general AI chat writes the answer that sounds most likely. It can mix a strong trial with a blog post, and it can quote a study that was never published. This tool works the other way round. It finds the studies first, throws out the weak ones, and only then lets the AI read what is left.
| Question | General AI chatbot | This fact-check |
|---|---|---|
| Where the answer comes from | What it learned in training or finds on the web. Blogs, news, forums and studies, all mixed together. | PubMed only, the US National Library of Medicine's index of peer-reviewed research. Human studies from the last 25 years. |
| Which studies count | No fixed rule you can see. A mouse study and a large trial can carry the same weight. | Ranked by design, strongest first. Only the top 12 papers per claim reach the AI. |
| Weak journals | No floor. | Only journals with an impact factor of 3 or more can be cited. A journal we cannot score is left out. |
| Made-up references | Can happen. A real-looking title, authors and year for a paper that does not exist. | Blocked by the server. Any citation that did not come back from the search is deleted. If nothing is left, the verdict is Not proven. |
| How strong the words are | Nothing stops it from calling a lab result "proven". | The words follow the evidence. Only pooled studies and controlled trials can say "reduces" or "increases". Observational studies only get "is linked to". |
| Can you check it? | Often no link, or a link that does not open the paper. | Every study shows its PMID and opens on PubMed, so you can read it yourself. |
How we rank the studies
Not all studies are equal. The order below is the one evidence-based medicine has used for decades, and the verdict always leans on the strongest study the search found.
- Many studies pooled together. Meta-analyses and systematic reviews. One study can be a fluke. Pooling several averages the flukes out.
- A controlled trial. People are split at random and only one thing changes. It is the only design that shows cause and effect.
- People observed, not tested. Cohort and survey studies. They show a link, not a cause: people who eat one way often live differently in other ways too.
- Lab, animal or expert opinion. Where ideas start, not where they are proven. These never carry a verdict on their own: a claim that rests only on a mouse study is Not proven.
Why a journal floor?
A journal's impact factor counts how often other scientists cite its papers. It is a rough signal, not a perfect one. It still clears out the long tail of journals with little real review, including the ones that publish almost anything for a fee.
The cost is that a good study in a small journal can be missed. We accept that, so a weak study never sneaks in as the proof.
What it does not do
- It reads each study's summary (the abstract), not the full paper.
- It can only judge what the search found. A study it missed does not count.
- It checks claims. It does not give medical advice, and medication questions stay with your prescriber.
Interactive guides that turn one set, one run or one morning check into the number you need next. Each one opens on guides.drbotezelli.com with the method and the evidence behind it.
The 1RM Predictor
Estimate your one-rep max from a set you have already done, with the right equation for your sex and training level.
Open the calculator → RPEThe RPE Method
Enter a set you finished and get the true percentage it was, and what to load next.
Open the calculator → ProteinYour protein target
Your daily protein target from your weight and how you train, divided across four meals.
Open the calculator → HydrationThe Exercise Hydration Guide
Fluid, sodium, carbohydrate and potassium for the session ahead, set by its duration and conditions.
Open the calculator → VO2 maxThe Cooper 12-Minute Test
Run for 12 minutes, enter the distance, and see where your result sits for your age and sex.
Open the calculator → LoadThe Training Load Tracker
Turn every workout into one number, then check your acute to chronic ratio against the safe zone.
Open the calculator → ReadinessThe Daily Readiness Score
A 30-second morning check on sleep, soreness, energy, mood and stress that scores how ready you are to train.
Open the calculator → Metabolic ageWhat is your metabolic age?
Four things you can measure at home, turned into the age your body is behaving like.
Open the calculator →Want someone to read the evidence for you?
A free 20-minute call to talk through where you are and whether a program fits. No pitch, no pressure.