Blog
Evidence-based writing on health data, wearables and what the research actually supports.
Do ACL injury prevention programs work?
ACL injury prevention programs were linked to about half the ACL rupture risk (pooled RR 0.46) in a 2026 meta-analysis of 25,166 athletes. What it shows.
Can training load ratios predict injury?
Acute:chronic workload ratio and injury: a 2026 meta-analysis of 16 studies (797 athletes) found g = 0.35, with very high heterogeneity and mixed evidence.
Apple Watch blood pressure alerts: who gets missed?
A JAMA modelling study estimated the Apple Watch hypertension alert would miss 59% of US adults with undiagnosed high blood pressure. What that means.
Artificial sweeteners and heart health: 37 studies pooled
A 2026 meta-analysis of 37 studies linked higher artificial sweetener intake to type 2 diabetes (HR 1.26). Why that link is hard to interpret.
Beetroot juice and performance: what 33 studies found
A 2026 meta-analysis of 33 studies in 519 athletes found small to moderate effects of beetroot juice on sprint, power and VO2 max. Here are the limits.
Caffeine dose and time trials: what 48 studies found
A 2025 network meta-analysis of 48 studies found low and moderate caffeine doses both shortened time trials, with no significant gap between them.
Is 90 g of carbohydrate an hour still the ceiling?
A 2026 review proposes raising the carbohydrate ceiling in endurance exercise from 90 to 120 g/h for trained athletes. What it does and does not show.
Coffee and AFib recurrence: what the DECAF trial found
In the DECAF trial, AF or flutter recurred in 47% of adults assigned to caffeinated coffee versus 64% assigned to abstain over six months. What it shows.
How often do blood biomarkers flag concussion?
In a study of 323 athletes with concussion, the blood marker GFAP was elevated in only 16.0% within 12 hours. What concussion blood biomarkers show.
Hamstring injury prevention: what 11 trials found
Hamstring injury prevention in a 2026 network meta-analysis of 11 trials and 9,282 football players: FIFA 11+ had a rate ratio of 0.55, certainty low.
Does remote monitoring help in heart failure?
A 2026 meta-analysis of 79 randomised trials found remote monitoring in heart failure cut all-cause mortality (RR 0.90). What counts as remote monitoring?
Drinking to thirst or to a plan: what ultra data show
In a 161 km ultramarathon, 67.0% of 383 starters drank to thirst and weight loss was similar across strategies. What the evidence on hydration shows.
What decides HYROX results? What the research shows
HYROX performance research: in 39,696 race results, running was about half of total time, and top-100 times improved about 19% (men) and 21% (women).
How common is low energy availability in athletes?
A 2025 meta-analysis of 59 studies classified 44.7% of 6,118 athletes as having low energy availability. What the pooled figure does and does not show.
Does the menstrual cycle change training performance?
Menstrual cycle and performance: a 2025 review of 19 hormone-verified studies (279 women) found 58% reported phase effects, but direction varied.
Do plant-based diets affect athletic performance?
Plant-based diet athletic performance: 8 trials found no significant strength difference, and a 10-study review pooled an aerobic effect size of 0.55.
Does a sauna after training improve endurance?
Does sauna after training improve endurance? A 2025 meta-analysis of 10 trials (199 people) found a trivial, imprecise effect and low certainty.
Does the timing of protein matter for muscle gain?
A 2025 meta-analysis of five trials found no difference in lean mass gains between protein before or after training (SMD -0.08). What it shows.
Do wearable readiness scores have evidence behind them?
Wearable readiness score validation: a 2025 review of 14 composite scores from 10 makers found little to no peer-reviewed validation. What has been tested.
Salt substitutes and heart health: what 34 trials show
A 2026 network meta-analysis of 34 trials estimated potassium salt substitutes lower systolic blood pressure by 4.4 to 4.6 mmHg. What the evidence covers.
Does shockwave therapy work for tendon pain?
In a 2026 meta-analysis of 8 trials (458 people), shockwave therapy did not differ significantly from sham for Achilles tendinopathy pain or function.
How accurate is a smartwatch at flagging sleep apnea?
In a 152-adult lab comparison, a Samsung smartwatch flagged moderate to severe sleep apnea with 94.1% sensitivity but 66.7% specificity. What that shows.
Strength before or after cardio? What reviews found
Strength before or after cardio? An umbrella review of 17 meta-analyses (1,492 people) found no significant effect of training order on adaptations.
Time-restricted eating and weight: what 20 trials found
A 2025 time-restricted eating meta-analysis of 20 trials pooled body weight 1.59 kg lower than unrestricted eating. What it found and left open.
How accurate are wearables at detecting ovulation?
Wearable ovulation detection accuracy: a 2026 meta-analysis of 27 studies pooled an accuracy of 0.88, with very low certainty. Who funds device studies.
Wearable step cadence during chemotherapy and survival
In 213 people on chemotherapy, faster peak walking cadence on a Fitbit was linked to lower mortality (HR 0.87 per 10 steps/min). Preliminary data.
How accurate is the VO2 max on your watch?
Wearable VO2 max accuracy varies: an independent test of 28 adults found a 13.3% average error on Apple Watch, and a 13-study review found mixed results.
Can wearable movement data detect early Parkinson's?
Accelerometer models for prodromal Parkinson's disease scored an AUPRC of 0.07 in UK Biobank, and performed worse in two independent cohorts in 2026.
How accurate is AI health advice? A 2026 study's answer
A 2026 Penn State study found 76.2% of health answers from four AI chatbots were accurate, as 230 million people a week ask ChatGPT about health.
Alcohol and health risk: what a 2026 modelling study found
A 2026 modelling study of 7,200+ articles found no protective effect of alcohol at any level and a 1 in 25 mortality risk at 14 drinks a week.
Alzheimer's blood test p-tau217: what a result can show
In about 2,700 cognitively healthy adults, very high p-tau217 was linked to a 78% chance of impairment within 10 years. What the test can show.
CGMs without diabetes: what the research shows so far
In 3,634 adults without diabetes, time in a tight glucose range showed moderate discrimination for predicted heart risk (AUC 0.75). What CGMs show.
Creatine and brain health: what the evidence supports
Searches for creatine and the brain rose 118% in a year. A 2026 review of 1,542 older adults found possible cognitive links, but low certainty.
Dietary fiber and health: what 185 studies found
A Lancet analysis of 185 observational studies and 58 trials linked 25 to 29 g of daily fibre to 15 to 30% lower mortality. What it supports.
Protein at 1.2 to 1.6 g/kg: what the evidence supports
The 2025 to 2030 US Dietary Guidelines set protein at 1.2 to 1.6 g/kg. A 49-trial meta-analysis found training gains plateau near 1.62 g/kg.
Early-onset colorectal cancer: what 40 years of data show
Swiss registry data on 96,410 cases show colorectal cancer rising about 0.5% a year in adults under 50, while falling at ages 50 to 74. The details.
Extreme heat and health: what the Lancet Countdown counts
The 2025 Lancet Countdown counted an average 546,000 heat-related deaths a year (2012 to 2021), up 23% per 100,000 since the 1990s. What it counts.
Menopause hormone therapy: what the FDA label change means
In November 2025 the FDA began removing boxed warnings on heart disease, breast cancer and dementia from menopause hormone therapy. What changed.
Microplastics in the body: the measurement debate
A 2026 study found plastic in 84% of heart attack patients' blood vs 32% with normal arteries (n=61). Association only, and methods are debated.
NHS-Galleri: what the multi-cancer blood test trial showed
NHS-Galleri tested a multi-cancer blood test in 142,942 adults. It missed its primary endpoint, yet stage IV diagnoses were about 14% lower.
Normal lab range vs your own baseline: why both matter
A blood result can be unusual for you yet sit inside the population range. Why within-person variation matters, with a creatinine example (4.3% vs 18.3%).
Gray market peptides: what the FDA panel reviewed
In July 2026 an FDA advisory panel backed six of seven gray market peptides for compounding, including BPC-157. What human data exists so far.
Why semaglutide failed in two Alzheimer's trials
The EVOKE trials of oral semaglutide in 3,808 people with early Alzheimer's showed no benefit over placebo at 104 weeks. What that tells us.
Sleep tracking and orthosomnia: what a 2026 survey found
In a 2026 US survey of 1,280 adults, 32.4% tracked sleep and 30.9% of trackers screened positive for orthosomnia risk. What that anxiety means.
Smartwatch AF screening: what the EQUAL trial found
In the EQUAL trial, smartwatch monitoring found new atrial fibrillation in 9.6% of high-risk over-65s vs 2.3% with standard care. What it shows.
Health claims on TikTok: what two audits found
Two peer-reviewed audits found misinformation in 31% of TikTok stroke videos and about 44% of sinusitis videos, mostly from non-professionals.
Ultra-processed foods: what the Lancet series says
A 2025 Lancet series drew on 100+ studies linking ultra-processed food to poor health across most organ systems. Where causation is still open.
Whole-body MRI screening: what the evidence supports
One review found 95% of whole-body MRI scans in people without symptoms were abnormal, yet 91% of lesions were benign. What the evidence shows.
7,000 steps, not 10,000
A 57-study meta-analysis found 7,000 daily steps associated with 47% lower all-cause mortality. Beyond that, the curve flattens for most outcomes.
Why biological age tests disagree
Epigenetic clocks are technically reliable (ICC above 0.90) but biologically unstable. A 2026 Aging Cell study found the two are almost entirely unrelated.
How much deep sleep you actually need
Across 169 studies, deep sleep averaged 20.4% of the night and changed little with age. Fragmentation, not collapsing deep sleep, is the real age signal.
The exercise trial with a drug-sized effect
In a phase 3 RCT of 889 colon cancer patients, structured exercise cut recurrence risk 28% and death risk 37% over a median 7.9 years of follow-up.
Your HRV number is not the point
In 1,175 healthy adults, mean rMSSD was 27.3ms with a standard deviation of 22.2ms. The spread between people is why HRV charts tell you so little.
What a rising resting heart rate tells you
Resting heart rate is the cheapest longitudinal biomarker you own. In a 5,262-person study, its drift from baseline flagged infection days before symptoms.
VO2 max, and the part of the curve you control
In 122,007 patients, elite cardiorespiratory fitness was associated with 80% lower mortality than low fitness. No upper limit of benefit was observed.
Your wearable can see illness coming
In 5,262 people, 81% of COVID cases showed wearable-detectable changes, a median of 4 days before symptoms. The signal exists. It has nowhere to go.
What the evidence actually says about Zone 2
A 2025 Sports Medicine review found most evidence argues against Zone 2 as optimal for mitochondrial capacity. Higher intensities performed better.
Forward Head Posture: 10 Habits Backed by Research
Exercise reliably changes craniovertebral angle across 7 RCTs and 627 people. 10 habits the research supports, and what the pain evidence really shows.
How to Fix Bad Sleep: 10 Habits With Real Evidence
Across 60,977 people, sleep regularity beat sleep duration as a mortality predictor. 10 sleep habits with real numbers behind them.
What Does a Dry Sauna Actually Do? The 20-Year Data
Over 20 years, sudden cardiac death hit 5.0% of frequent sauna users vs 10.1% of weekly users. What the Finnish data shows, and its limits.
Why Am I Always Tired? What the Data Says
14% of US adults have iron deficiency and most would never be tested. Three measurable, commonly missed contributors to persistent tiredness.
A Morning Routine for Longevity, Built on Three Findings
The brightest tenth of daytime light exposure tracked with 17-34% lower mortality across 88,905 people. A morning routine built on three findings.
How to Build an Anti-Aging Lunch: What 105,000 People Ate
Top-quintile diet quality tracked with 86% higher odds of healthy aging at 70 across 105,015 people. What that looks like as an actual lunch.
An Evening Routine for Longevity: What the Light Data Shows
Brightest-tenth night light exposure tracked with 21-34% higher mortality across 88,905 people. An evening routine built on the light and meal data.
Recovery for 5K and 10K Runners: What the Meta-Analyses Rank
Across 99 studies, massage beat ice baths 5 to 1 for soreness and stretching did nothing. What recovery methods rank for 5K and 10K runners.
The Best Exercise to Live Longer, According to 147,374 People
Across 147,374 people over 30 years, combining aerobic and resistance training tracked with 45-58% lower mortality. The dose numbers for each.
How to Improve Your HRV: What Actually Moves the Number
Exercise raised RMSSD with a pooled effect size of 0.84 across 16 trials. What moves HRV, what suppresses it, and why your number is not comparable.
How to Improve Your Running: Three Levers With Evidence
Strength training improves running economy and polarized training improves VO2peak, but neither does what most runners assume. The evidence, with effect sizes.
Ice Plunges: What the Randomized Evidence Actually Shows
A meta-analysis of 11 trials and 3,177 adults found cold plunges cut stress at 12 hours, raised inflammation acutely, and did nothing measurable for immunity.
Air Quality and Blood Pressure: How Big Is the Effect?
Ten randomized trials found air purifiers cut systolic blood pressure by about 4 mmHg. The PM2.5-to-blood-pressure evidence, with effect sizes and limits.
How to Improve Home Air Quality: What the Measurements Show
Gas stoves raise indoor NO2 by 4.0 ppbv and it reaches the bedroom. Four home air quality interventions ranked by what randomized trials actually measured.
Plasticizers and Health: What the Evidence Supports
A global model attributed 356,238 deaths to DEHP exposure in 2018. What the phthalate and BPA evidence supports, and how much is modelled rather than measured.
How to Reduce Plasticizers at Home: The Three-Day Experiment
Three days of fresh, unpackaged food cut urinary BPA by 66% and DEHP metabolites by over half in 20 people. What that study tested, and what it did not.
Heart Disease Prevention: Which Tests, and When
The 2026 ACC/AHA guideline added lifetime Lp(a) testing and lipid screening from 19. Which tests are recommended, at what age, and why it changed.
How Much of GLP-1 Weight Loss Is Muscle?
Across trials, lean mass accounts for anywhere from under 15% to 60% of weight lost on GLP-1 therapies. That spread is the story.
Semaglutide's Heart Benefit Did Not Track Weight Loss
In SELECT, 17,604 adults on semaglutide saw fewer cardiac events regardless of how much weight they lost. Why the scale is a poor risk readout.