AI health research
Your health
questions deserve
better evidence.
Ask anything. Minkey searches the research, separates signal from noise and shows you exactly what the evidence can — and cannot — tell you.
Open questions about how something works or whether it helps.
01
Ask anything
Turn confusing health questions into clear research questions — then see them answered from published work rather than opinion.
02
See the evidence
Understand the studies behind a conclusion: their design, size, quality, limitations and the uncertainty that remains.
03
Check the source
Trace every statement back to a real scientific publication, with its metadata, funding disclosures and a link to the original.
Health information is everywhere. Context isn’t.
A headline compresses a study into six words. A post quotes one result and drops the population it came from. A single trial in eleven people becomes a rule that applies to everyone.
None of that is dishonest on its own — it is what happens when research travels. Minkey puts the context back: what was measured, in whom, for how long, and how confident anyone can reasonably be.
Example analysis
Real sources, saved so you can look before you run one
Does walking after a meal reduce the post-meal rise in blood glucose?
Evidence posture
Mostly supported
The weight of the research leans this way, with some caveats or inconsistent results.
Evidence strengthHow good the underlying body of research is — study design, size and consistency — regardless of which way it points.
Moderate
Reasonable human research, with limits on size, length or design.
Research confidenceMinkey’s confidence that this summary fairly reflects the sources it retrieved. It is not a truth score.
78%
How well this synthesis represents the retrieved research — not the probability that the claim is true.
Bottom line
Short, easy walking soon after eating consistently lowers the glucose rise that follows a meal, and the effect appears within minutes rather than hours. The size of that effect is modest, and it is measured over a few hours in small crossover trials of roughly 10 to 30 people. Whether repeated post-meal walks translate into better long-term glycaemic control or lower disease risk has not been established by this literature.
What the research suggests
Pooled analyses of randomised crossover trials find that interrupting sitting with brief activity lowers both post-meal glucose and insulin, with walking the most effective mode tested.
Timing appears to matter more than duration: a 10-minute walk taken immediately after a glucose load lowered peak glucose, while a longer walk started 30 minutes later did not.
Frequent short interruptions — roughly every 15 to 30 minutes — produced larger reductions in subgroup analyses than fewer, longer bouts.
The effect survives outside the laboratory: replacing sitting with standing and walking lowered waking-hour glucose in free-living conditions, though 24-hour measures did not shift.
Evidence map
Every source Minkey used, grouped by what kind of evidence it is. Circle size follows the reported sample size; the outline follows the study design.
Systematic reviews & meta-analyses
2 sources
Systematic review & meta-analysis
The Acute Effects of Interrupting Prolonged Sitting With Regular Activity Breaks on Postprandial Glucose and Insulin in Adults: A Systematic Review and Meta-Analysis
Gale et al. · Obesity reviews : an official journal of the International Association for the Study of Obesity · 2026 · Not reported
Systematic review & meta-analysis
Acute effects of exercise snacks on postprandial glucose and insulin metabolism in adults with obesity: a systematic review and meta-analysis
Chang et al. · Frontiers in nutrition · 2025 · Not reported
Randomised controlled trials
5 sources
Randomised crossover trial
Positive impact of a 10-min walk immediately after glucose intake on postprandial glucose levels
Hashimoto et al. · Scientific reports · 2025 · 12 participants
Randomised crossover trial
Modifying the timing of breakfast improves postprandial glycaemia in people with type 2 diabetes: A randomised controlled trial
Bravo-Garcia et al. · Diabetes & metabolic syndrome · 2024 · 11 participants
Clinical trial
Substituting sitting with standing and walking in free-living conditions improves daily glucose concentrations in South Asian adults living with overweight/obesity
Dey et al. · European journal of applied physiology · 2026 · 14 participants
Randomised crossover trial
One minute of light-intensity stair-stepping decreases postprandial glycaemia in the evening in non-diabetic adults: A randomized controlled trial
Morales et al. · Experimental physiology · 2026 · 30 participants
Randomised controlled trial
Interrupted sitting at work can alter postprandial glucose, triglycerides and non-esterified fatty acids (NEFA)
Mat Azmi et al. · The Medical journal of Malaysia · 2025 · 24 participants
This example was written from the abstracts of the seven sources above. Running it live searches the literature again, right now.
Source DNA
A study is more than its headline.
Two papers can reach the same conclusion from completely different evidence — a six-week trial in twelve people, or a pooled analysis of thirty-nine. Minkey reads the metadata behind each source and shows you what kind of study you are actually leaning on.
The Acute Effects of Interrupting Prolonged Sitting With Regular Activity Breaks on Postprandial Glucose and Insulin in Adults: A Systematic Review and Meta-Analysis
Gale et al. · Obesity reviews : an official journal of the International Association for the Study of Obesity · 2026 · DOI 10.1111/obr.70152
- Study design
- Systematic review & meta-analysis
- Participants
- Not reported
- Subjects
- Human
- Publication
- Obesity reviews : an official journal of the International Association for the Study of Obesity
- Year
- 2026
- Follow-up
- Not reported
- Peer reviewed
- Yes
- Open access
- No
Independence lensFunding does not determine whether a study is valid. Minkey displays it as context, not a verdict.
No funding information found
Compare
Compare ideas, not opinions.
Put two options against the same literature and see where one actually pulls ahead — and where the honest answer is that they are the same.
On the outcomes people usually care about — body composition, metabolic markers, sleep — pooled analyses find no reliable difference between morning and evening training. Where differences do appear they are small, inconsistent and mostly acute. The clearest signal in this literature is not morning versus evening at all: training at a time that matches your chronotype and your schedule is what people sustain, and consistency is what the outcomes track.
Body composition
Moderate strength
Morning exercise
No significant advantage for morning training across 11 studies reporting anthropometric outcomes.
Evening exercise
No significant advantage for evening training in the same pooled analysis.
Metabolic markers
Low strength
Morning exercise
Repeated morning exercise was associated with fasting glucose 0.25 mmol/L higher than afternoon or evening training in a meta-analysis of 20 studies.
Evening exercise
Slightly lower fasting glucose with afternoon or evening training, though heterogeneity was high and confidence intervals wide.
Performance
Low strength
Morning exercise
Acute neuromuscular performance is generally lower in the morning, particularly for power and strength tasks.
Evening exercise
Late afternoon and early evening tend to show better acute performance, but chronic training studies find no universally optimal time.
Sleep
Very low strength
Morning exercise
Morning training is often assumed to protect sleep, but the pooled sleep outcomes showed no significant timing effect.
Evening exercise
Evening aerobic exercise did not reliably impair sleep; one uncontrolled eight-week programme was followed by improved sleep quality scores.
Personal fit and adherence
Low strength
Morning exercise
Best for morning chronotypes: aligning training with chronotype produced larger blood pressure and sleep improvements than misaligned training.
Evening exercise
Best for evening chronotypes, for the same reason. The trial tested alignment, not clock time.
How it works
01Ask
Enter a question, a claim, an article or a comparison. No account, no setup, no onboarding.
02Research
Minkey turns it into a scientific search, queries Europe PMC, Crossref and ClinicalTrials.gov, and screens what comes back for design, relevance and quality.
03Understand
A reasoning model reads only the retrieved sources and explains what they show — preserving uncertainty, disagreement and every citation.
Pay for research, not another subscription.
Crypto payments · coming soonPay as you go
Coming soonWhen the free runs are not enough, buy research by the run. No plan, no renewal date, no card on file — settle in USDC on Solana and the run starts immediately.
Quick research
One question, one evidence map.
$0.25 USDC
Deep evidence scan
Wider retrieval and a closer read of the literature.
$0.75 USDC
Compare
Two options, matched against the same literature.
$0.40 USDC
Prices are indicative while the payment rail is being built. Nothing is charged today, no wallet is connected, and the free runs above need no account either way.
Built around a simple principle
Better health decisions begin with better information.
Dr. Ben Tapper
Founder
Dr. Ben Tapper has spent years communicating about health, wellness and scientific evidence. Minkey turns that philosophy into a tool anyone can use.
@DrBenTapper1