Where the numbers come from
Each line or label on this site has a paper or a guideline behind it. They are listed here with what the site does with each one. A number that is only an estimate says so on its own page.
What is on the page, and where it is from
| On the site | Where it is from |
|---|---|
| 0.8 m/s line on the speed log chart, and a 4-metre walk slower than 5 seconds flagged for a clinician to look at | BC Guidelines, Frailty in Older Adults (2017), gait speed sheet. See reference 1. |
| Faster walking, longer survival: about 12% lower mortality for each 0.1 m/s | Studenski et al., JAMA 2011, 34,485 adults 65 and older. See reference 2. |
| 0.05 m/s small change, 0.10 m/s substantial change on the trend labels | Perera et al., J Am Geriatr Soc 2006, measured between timed walks. See reference 3. Applying the same labels to the Gait Check's 90-day slope of phone data is this site's own choice. |
| Two trials, the faster counts, usual pace, on the 4-metre walk | NIA Short Physical Performance Battery protocol. See reference 4. |
| Speed bands (under 0.6, 0.6 to 1.0, 1.0 to 1.2, 1.2 and over) | The gait speed literature as summarised in Fritz and Lusardi (2009). See reference 5. Used as plain labels, not as a diagnosis. |
| An improving speed goes with better survival | Hardy et al., J Am Geriatr Soc 2007. See reference 6. |
| Timed Up and Go 12 second screening line | Podsiadlo and Richardson (1991), and the CDC STEADI program. Reference values: Bohannon (2006). See references 8 to 10. |
| 30-second chair stand protocol and age-60-to-94 reference norms | Rikli and Jones (1999), and the CDC STEADI 30-second chair stand test. See references 7 and 10. |
| Five-minute check lines: the chair-stand “below average” counts by age (60 to 94) and sex, and the tandem stand held for 10 seconds | CDC STEADI 30-second chair stand sheet (2017) and CDC STEADI 4-stage balance test sheet. See references 7 and 15. The walk uses the BC 0.8 m/s line (reference 1). |
| Five-minute check, the newer signs: the one-leg stand held 10 seconds; grip (a screen question, and a typed reading against 27 kg for men and 16 kg for women); company (the three-item loneliness scale, 6-7 Watch, 8-9 Act); steps (6,000 a day at 60 and over, 8,000 under 60) and sitting (over 8 hours); three brisk 1-2 minute bursts a day | Araujo (2022), Leong (2015), Cruz-Jentoft / EWGSOP2 (2019), Hughes (2004), Paluch (2022), Ekelund (2016) and Stamatakis (2022). See references 22 to 28. Each is a group finding read as Good, Watch or Act; none is a forecast for one person. |
| The phone’s own walking data read on the Gait Check: steps, speed, asymmetry, double support and step length from an Apple Health export | Studies from the University of Maryland trauma centre, in adults after an operation for a leg or pelvis fracture. The phone’s step count tracked how people rated their own physical function, and people whose fracture did not heal regained steps more slowly (reference 12). More daily steps, a faster walk and less asymmetry before the injury went with going home from hospital (reference 13). Recovery curves for the five measures: reference 14. These show the numbers move with recovery; they do not predict what will happen to one person. |
| Walking dashboard: everyday iPhone speed called “possibly” changed from 0.10 m/s and “likely” from 0.14 m/s along a 12-week line | Apple’s own validation (179 people): one person’s iPhone speed has a median minimal detectable change of 0.14 m/s (range 0.08 to 0.23). See reference 16. Below that, a change may be the phone, not the person. |
| Everyday speed and timed-walk speed kept apart | Speed in daily life runs about 0.22 m/s below supervised speed in healthy older adults (meta-analysis of 11 studies). See reference 17. So the 0.8 m/s line is used only for timed walks. |
| Speed and step length trusted more than double support | Against lab body sensors, the iPhone was valid and reliable for speed and for step length in adults and seniors, but poor to moderate for double support. See reference 18. |
| Steps: trust the direction, not the exact count; never added across phone and watch | Phone apps miscounted daily steps by about 30% against a thigh sensor, an Apple Watch by about 6% (reference 19). Wrist trackers in older adults missed about 21% of steps at 1.5 km/h, 45% at 1 km/h, and more with a walking aid (reference 20). |
| Life-space score (0 to 120) on the dashboard | The UAB Life-Space Assessment, Baker, Bodner and Allman (2003). See reference 21. |
| Exercise to prevent falls, adults 65 and older at increased risk | US Preventive Services Task Force, Grade B. See reference 11. |
The papers and guidelines
What these numbers are, and are not
Walking speed is a screening signal that tracks risk across large groups. It does not diagnose anything and it cannot predict what will happen to one person. A slower walk is a reason to talk to a clinician, and a faster one is a reason to keep going. The fracture studies (references 12 to 14) come from one trauma centre and from people who own an iPhone and carried it. Phones that stay on a desk undercount steps.
TODAY: The 4-metre timed walk (/speed), the Gait Check (/check), the 20-second phone walk test (/walk-test), and the 30-second sit-to-stand (/chair-stand) run in your browser: nothing is uploaded, there is no account, and nothing is billed. NEXT: The clinician-review trajectory draft path (where AI drafts a trend summary from on-device metrics and a licensed clinician signs or declines it) is in design and not live.
The phone walk test and the Gait Check give estimates, and each page says which. When a number on a page has no source here, tell us and we will fix the page. Last reviewed 2026-10-14.