Research Decoder
Understand the research without becoming a researcher.
Paste an abstract, excerpt, or research paper. Research Decoder explains what the researchers found, what they actually did, what the numbers mean, and what the study can—and cannot—tell you.
See what Research Decoder gives you
Real output from an actual run on the abstract below. Part of the result is shown; nothing is reworded.
The abstract — “Association between ultra-processed food intake and all-cause mortality: prospective cohort study”
In this prospective cohort of 44,551 French adults aged 45 years and older followed for a median of 7.1 years, ultra-processed food represented 14.4% of the weight of total food consumed and 29.1% of total energy intake. Ultra-processed food consumption was associated with younger age, lower income, lower educational level, living alone, higher body mass index, and lower physical activity level. An absolute increase of 10% in the proportion of ultra-processed food in the diet was associated with a 14% higher risk of all-cause mortality (hazard ratio 1.14, 95% CI 1.04 to 1.27) after adjustment for a range of confounding factors. Sensitivity analyses excluding participants who died within the first two years of follow-up did not materially change the estimates. Limitations include self-reported dietary data and the observational design, which precludes causal inference.
Real run, 2026-10-08 — one of the tool's own built-in examples.
The finding
In a study of over 44,000 French adults, each 10-percentage-point increase in the share of ultra-processed food in the diet was associated with a 14% higher hazard of dying from any cause over a median follow-up of about seven years. The researchers found this association after adjusting for a range of confounding factors, but the observational design means causation cannot be inferred.
What the main number means
- Each 10-point rise in the share of ultra-processed food in the diet was associated with a 14% higher rate of death during follow-up, relative to those with a 10 points lower share. This is a relative comparison, not an absolute personal risk figure.
- A hazard ratio is not the same as personal risk. To know what this means in absolute terms for an individual, you would need the baseline mortality rate in the study population, which is not reported in this excerpt.
What it doesn't establish
- Causation: the researchers explicitly state the observational design precludes causal inference
- What the absolute increase in mortality risk was for any individual, because the hazard ratio alone does not convey that without baseline risk data not provided in this excerpt
- Whether the association holds in populations that differ from this sample in age, nationality, dietary culture, or other characteristics
- Which specific foods within the ultra-processed category, or which properties of those foods, might drive the association — the excerpt does not address this
In a Nutshell
- When
- You read a research finding — a paper, an abstract, or a headline about one — and want to know what it actually says.
- You give
- The abstract, excerpt, or paper text — or upload it. A headline or a second paper, if you're checking coverage or comparing studies.
- You get
- The actual finding, what the study did, what the numbers mean, and what it does and doesn't support — without inventing outside research or a confidence label it can't back up.
- The edge
- It says outright when you've only given it an abstract, not the full paper, and when a question needs outside research it doesn't have.
Good to Know
- Only paste what you actually have — an abstract works fine, and the tool says so rather than treating it as the full paper
- Headline Check needs the underlying research, not just the headline — without the paper, there's no way to tell whether the coverage matches it
- 'What does this mean for me?' only appears after a decode, and it grounds every answer in the actual paper text plus what you say about your situation — never a generic risk score
- The tool says plainly when a question needs outside research it doesn't have, rather than inventing a scientific consensus to sound complete
Questions people ask
What does a hazard ratio of 1.14 mean?
The event happened at a 14% higher rate in one group than the comparison group during the study. It's a relative figure: without the baseline rate, it doesn't tell you any one person's actual risk.
Can an observational study prove cause and effect?
No. It can show an association that holds after adjusting for known factors, but something unmeasured could explain part of it. Randomized trials are how cause is usually tested.
Can I upload a full paper?
Yes — paste text or upload a PDF. If you only give it an abstract, it says so, because limitations and methods often aren't in the abstract.