Disclosure: This article contains affiliate links. If you buy through them we may earn a commission, at no extra cost to you. It does not change what we recommend.
Can a $249 Smart Scale Replace My DEXA Scans? I Read the Study Behind Hume’s “98% Accurate” Claim.
Instagram told me the Hume Pod was “98% as accurate as DEXA.” I’ve paid for DEXA scans twice a year for years, so I went looking for the evidence — and found an independent Texas Tech study that changed what I bought.
Straight up: I chose and bought the Hume Pod 2.0 after the research below. I haven’t weighed in on it yet, so nothing here is my own measurement data — that comes in the follow-up, alongside my next DEXA. This article contains affiliate links; The Niftiest may earn a commission if you buy through them, at no cost to you. That didn’t change the research or the conclusion. And to be clear about what this is not: a smart scale is not a substitute for a clinically ordered DEXA for bone density. Mine were never about bone.
The verdict up front
I’m replacing most of my routine DEXA scans with a Hume Pod 2.0 — not because I believe it produces the same numbers as DEXA, but because independent testing suggests it’s accurate and, more importantly, repeatable enough for the thing I actually track: whether my body fat is going down while my muscle holds or grows. Three things to know before you buy one for the same reason:
- “98% DEXA accuracy” is marketing shorthand, not what the study says. The independent study found Hume tracks DEXA very closely (r = 0.94) but was not statistically equivalent to it for body-fat percentage, with a mean error of about 3 points.
- The independent study tested the original Pod, not the Pod 2.0 I bought. Hume says 2.0 is “rebuilt for greater accuracy.” That is the company’s claim, and I’m taking it as one.
- Nobody has shown yet that it tracks changes over months the way DEXA does. The researchers said so themselves. That’s exactly how I plan to use it, so I’m keeping one DEXA to check.
Hume Pod 2.0 — $249 at humehealth.com →
Price as shown on Hume’s US product page on September 24, 2026 (listed as reduced from $383). The page’s accuracy wording that day read “99% body-fat accuracy compared with DEXA” and “±2% from a DEXA scan”; the ads I saw said 98%. Hume’s own materials have used several formulations. Treat all of them as claims.
Why I’ve been paying for DEXA
For a few years I’ve had a DEXA scan once or twice a year. Not for bone density — for body composition. I don’t just want to know whether my weight moved; I want to know whether I’m losing fat while keeping or adding muscle, because plenty of “successful” weight loss is the wrong tissue leaving. DEXA gives an excellent snapshot. It is also inconvenient and expensive enough that I only ever get two data points a year, which is a lousy way to see a trend.
Then Instagram served me an ad: a $250 scale claiming 98% DEXA accuracy. My honest reaction was “that can’t be true.” My second reaction was that if it were even half true, it would change how I track this. So I went looking.
The question I was actually asking
This was never a smart-scale roundup. I didn’t compare Hume against Withings, Garmin, Renpho or InBody, and this isn’t pretending to. The decision was narrow: keep doing DEXA, or buy a Hume Pod. The question underneath it: is Hume accurate enough for my purpose that I can stop paying for scans so often?
What Hume claims, and how the Pod works
Most smart scales are foot-only bioelectrical impedance (BIA): a small current runs up one leg and down the other and the scale estimates the rest of you. The Hume Pod adds hand contacts, so current runs through arms, legs and trunk, measured segmentally at multiple frequencies. That architecture is why it’s plausible for a consumer device to do better than a bathroom scale.
Hume advertises the Pod 2.0 with about 45 metrics — body fat, lean and muscle mass, visceral fat, hydration, body symmetry — plus a 2.8-inch display, a rechargeable battery it rates at up to a year per charge, a 441-lb maximum, Apple Health and Google Fit sync, and an app “Digital Twin” body model. On accuracy, the product page says “99% body-fat accuracy compared with DEXA” and “±2% from a DEXA scan,” citing an independent Socotech evaluation. Elsewhere the company has used 98% and ±3%. When the number keeps changing, it’s a claim, not a measurement.
Then I found an actual independent study
In July 2026, Grant Tinsley’s Energy Balance & Body Composition Laboratory at Texas Tech posted Validity and Test-Retest Reliability of the Hume Pod Bioimpedance Analyzer for Body Composition Assessment (medRxiv, DOI 10.64898/2026.07.17.26358337). It’s a preprint — not yet peer-reviewed — but it’s a real lab, a real protocol, and a real reference standard.
Why it’s worth more than a testimonial: 67 adults (42 women, 25 men, average age 37, average BMI 24.6); Hume compared against both DEXA and a four-compartment (4C) body-composition model, which is closer to a true reference than DEXA alone; the lab bought its Pod from Hume in January 2026; Hume didn’t fund the work; and testing was standardized — restrictions on food, fluids, caffeine, nicotine, supplements and exercise, bladder emptied first.
What the study found
| Finding | Number | What it means for a trend-tracker |
|---|---|---|
| Test-retest reliability | ICC ≥ 0.993; technical error ~0.8 percentage points body fat, ~0.6 kg fat mass, ~0.6 kg fat-free mass | Step on twice, get nearly the same answer. This is the number I care about most. |
| Hume vs DEXA, body fat | Hume 27.1% vs DEXA 28.9% average; r = 0.94; mean absolute error 3.0 points; not statistically equivalent | Tracks closely, reads low. A single Hume number is not a DEXA number. |
| Hume vs DEXA, fat mass / fat-free mass | Fat mass not equivalent; fat-free mass met the equivalence criterion | The muscle side of my question held up better than the fat side. |
| Hume vs 4C model, body fat | Hume 27.1% vs 4C 26.4%; r = 0.95; MAE 2.6 points; equivalent for body fat, fat mass and fat-free mass | Against the better reference, it passed. |
| Limits of agreement vs 4C | SEE ~3.1 points; 95% limits ~±6.1 points | Individual readings can still be several points off. Don’t react to one. |
The authors wrote that the Pod’s cross-sectional validity compared favorably with numerous consumer BIA analyzers their lab had tested, naming devices like the InBody H2ON and Tanita BC-568 that had been among the better ones. That was the single most persuasive sentence in the whole decision.
The catch: correlation is not accuracy
Here’s where the “98%” needs qualifying. Two instruments can move in lockstep and still disagree about the absolute number. r = 0.94 does not mean “94% accurate,” and no marketing percentage means “if DEXA says 20%, Hume reads between 19.6 and 20.4.” The independent data shows Hume reading roughly two points lower than DEXA on average, with a typical miss of three points either way. That is very good for a consumer device. It is not DEXA.
And it’s why the study changed my mind in a specific way. Before, I read “98%” as marketing. After, my conclusion split in two: the marketing oversimplifies the accuracy, and the device underneath appears to be legitimately good. Those are different conclusions, and only the second one matters for what I need.
The bigger limitation
The study established validity at one point in time and repeatability within a visit. It did not establish that a Hume reading falling from 20% to 17% over six months means DEXA would show the same three-point decline. The researchers said additional work is warranted on longitudinal validity. Tracking over months is precisely how I intend to use it, so I’m not burying that. It’s the reason I’m keeping a DEXA on the calendar.
The other practical lesson from the protocol: participants didn’t hop on after breakfast. BIA is sensitive to hydration and recent activity. So my home rule is fixed conditions — first thing in the morning, after the bathroom, before food, coffee or exercise — and weekly or monthly trends, never a single reading.
Then Hume released the Pod 2.0
I’d researched the original Pod. In September 2026 Hume announced the Pod 2.0, “rebuilt for greater accuracy,” with the same hand-and-foot segmental design. If I was buying now, I wanted the current hardware for a device I expect to use for years. But be precise about what that means: Texas Tech did not test the Pod 2.0. Their unit was bought in January 2026; 2.0 didn’t exist. Hume says 2.0’s algorithm went through independent clinical testing and cites Socotech results against DEXA. I found no academic validation of 2.0 comparable to the Texas Tech work. So my reasoning for 2.0 is a buying judgment, not scientific proof: the original architecture tested unusually well, 2.0 keeps that architecture, the company says it improved it, and I’d rather own the newest version than save a little on the outgoing one.
What it can and can’t replace for me
Can largely replace
Routine body-fat trend checks; fat-mass and lean-mass trends; seeing whether weight loss is mostly fat; frequent feedback between the big milestones.
Cannot replace
A clinically indicated bone-density DEXA; any osteoporosis question; medical measurement of bone mineral density; laboratory-grade certainty; any diagnosis or treatment decision. Its “bone” metric is not DEXA bone density.
My plan, and the experiment I owe you
Not DEXA forever or Hume forever — Hume often, DEXA occasionally as a check. I’ll measure daily under fixed conditions and look at rolling trends. I’m keeping my next DEXA appointment and taking a Hume reading under the same conditions as close to it as I can. Then I compare body fat, fat mass and lean mass and learn my personal offset. If DEXA says 18.0% and Hume consistently says 16.5%, I don’t care that Hume is 1.5 low. I care whether 16.5 → 15.5 → 14.5 on Hume lines up with real fat loss while lean mass holds. If it does, most of my future DEXA appointments go away.
The money is the easy part of the argument. Each DEXA scan has cost me $120, and I’ve been getting two a year — $240 a year for two data points. The Pod 2.0 is $249 once for the basic body-composition readings (Hume also sells optional premium app features I haven’t bought), and it produces a data point every morning I want one.
| Horizon | Two DEXA scans a year at $120 | Hume Pod 2.0 | Difference |
|---|---|---|---|
| 3 years | $720 | $249 | $471 saved |
| 5 years | $1,200 | $249 | $951 saved |
| 10 years | $2,400 | $249 | $2,151 saved |
Assumes the Pod replaces every routine scan and lasts the period, and ignores any premium subscription. If I keep one $120 DEXA a year as a check — which is my actual plan to start — the saving is $120 a year, so the device pays for itself in just over two years and the real win is going from two readings a year to as many as I want.
The decision in one paragraph
I bought the Hume Pod 2.0 because I think it can replace most of the DEXA scans I’ve been using simply to watch fat go down and muscle stay up. I did not conclude it’s 98% interchangeable with DEXA, and I’d push back on anyone who tells you it is. The independent evidence on the original Pod shows very good repeatability, strong body-composition performance for a consumer BIA device, tight correlation with laboratory methods, meaningful individual error versus DEXA, and enough uncertainty that absolute readings shouldn’t be treated like clinical ones. For my use, a slightly imperfect measurement I can take every morning is worth more than a somewhat better one I get twice a year. Check back in a few months for the numbers.
Sources
- Tinsley GM et al., Validity and Test-Retest Reliability of the Hume Pod Bioimpedance Analyzer for Body Composition Assessment, medRxiv preprint, posted July 20, 2026, DOI 10.64898/2026.07.17.26358337 — all study figures above.
- Hume Health: Hume Pod 2.0 product page (price, claims, specifications, read September 24, 2026); original Hume Pod page; Hume’s smart-scale accuracy article.
- Systematic review and meta-analysis comparing BIA and DXA (PubMed 36853902) and a critical overview of BIA versus DXA (PMC6560329) — background on the method’s known limits.