I used your app with the [1] Nexpure CF586BLE (E8:6B:EA:82:B3:02) and [2] Hume BodyPod (A0:B7:65:95:11:BA) BIA scales. Both synced and paired easily with my laptop (Windows 11 v25H2). Note that the Nexpure CF586BLE is a rebranded Unique CF597BLE scale, made by Shenzhen Unique Scales Co. Ltd, China. The Unique Health app is made by Shenzhen Lefu Scale Co. Ltd, China.
BMI is an outdated metric and is total BS. I'd encourage you to drop its usage.
- Historical data is stored in metric units even if imperial units are selected in Settings (FAIL, please fix).
- Does not pull in historical data from either Hume or Unique apps (FAIL).
- Historical and export time stamps are not local time (FAIL).
- Not all data is exported.
- I didn't see the heart rate data in the metrics (FAIL).
- Body Wt is very close (PASS).
- The subcutaneous and visceral fat metrics are high (FAIL) compared to the Hume Health (v9.17.2.5237) and the Unique Health (athlete mode, v3.5.1.10) apps.
- Hume has a known TBW issue in its algorithm.
I would encourage you to verify your equations and conversions. I reviewed your "BLE Body Composition Scale — Data Retrieval & Decoding Protocol" and found several suspect conversions of the impedance data.
Go here for a Glossary of BIA terms
Go here for BIA Governing Equations
Here are a few comparisons between Blue2Scale, HumeHealth, and Unique Health apps with the Hume and Nexpure scales:
Wt (lbs): B2S Hume = 172.9; B2S Unique = 177.3;
Hume = 172.7; Unique = 177.4
BFM (%): B2S Hume = 27.0; B2S Unique = 28.9;
Hume = 21.2; Unique = 24.0
SMM (lbs): B2S Hume = 79.6; B2S Unique = 79.4;
Hume = 84.2; Unique = 76.2
TBW (%): B2S Hume = 52.0; B2S Unique = 53.3;
Hume = 44.5*; Unique = 55.7
BMR (kcal): B2S Hume = 1582; B2S Unique = 1583;
Hume = 1609; Unique = 1689
VFI: B2S Hume = 12; B2S Unique = 11;
Hume = 8; Unique = 9
SFM (%): B2S Hume = 25.1; B2S Unique = 23.5;
Hume = 18.5; Unique = 20.8
In general, I think your work is brilliant! Since you built this BLE from scratch, I'm guessing that you can add other meaningful BCA metrics that other apps leave out. Namely,
ICW/ECW (cellular health)
ECW/TBW (water balance)
ICW/SLM (muscle quality)
TBW/SLM (muscle hydration)
FFM = Wt - BFM = SLM + (MC + IS) = BCM + ECM
PM = FFM - (TBW + MC + IS)
SLM= TBW + PM
DLM = PM + MC
BCM = ICW + PM
ECM = ECW + MC + IS
LBM = SLM + MC
MA = (10Wt + 6.25Ht + 5 - BMR) / 5
BMR = 370+21.6*SLM / 2.20462
WHR = WC/HC
ASLM = SLM(R-arm) + SLM(L-arm) + SLM(R-leg) + SLM(L-leg)
ASLMI = ASLM/Ht^2
FFMI = FFM/Ht^2
BFMI = BFM/Ht^2
Your impedance package suggests 2 frequencies, 50 and 250 kHz. The 2024 Hume data states the BodyPod 5 and 50 kHz.; whereas, the 2026 Hume data states 20 and 100 kHz. Can you tell from the BE packets the actual impedance frequency? Or are you guessing at low and high?
Given what you know from the impedance data packet, can you calculate the segmental resistance and phase angles, and the whole body phase angle? If so, you can turn Blue2Scale into a very powerful app. The biggest hurdle for app accuracy is validation. Since the hardware is a fixed variable, the software is where you can shine.
- Direct Segmental Analysis: 8-electrode (4 foot, 4 hand) configuration for independent impedance measurements per segment.
- Multi-Frequency BIA: Multiple frequencies (e.g., 1–3000 kHz) with the specific breakdown (2+ below 50 kHz, 50 kHz for peak PhA, 2+ above 50 kHz)
- Validation: High correlation with gold standards (e.g., MRI, 4C model, DXA)
Absolutely essential. BIA is an indirect method, so device-specific validation studies (not just manufacturer claims) are what separate hype from reality. High ICCs (often >0.90 for lean mass, fat mass, and % body fat) with DXA or 4-compartment (4C) models are the benchmark for DSMF-BIA devices. Many peer-reviewed papers on 8-electrode multi-frequency systems show excellent agreement for whole-body and segmental measures in diverse populations, though limits of agreement are tighter for lean mass than fat percentage in some cases. Bonus points if validation covers varied groups (age, BMI, ethnicity, clinical conditions like obesity or diabetes). This criterion rightly emphasizes correlation plus low bias—raw numbers like ICC alone aren’t enough; Bland-Altman plots matter for real-world agreement.
- Precision and Reproducibility: Minimal variability in repeated measurements, clinical-grade accuracy
This is the practical clincher. Even the best tech fails without low test-retest variability (e.g., <1–2% CV for key metrics under standardized conditions). Clinical-grade DSMF-BIA devices excel here when protocols control for hydration, posture, fasting, exercise, and time of day. High reproducibility makes them useful for tracking changes over time (diet, training, recovery), not just one-off snapshots.
If you are interested in collaborating, I know nothing about programming in BLE, but I have a Master's in engineering and have done extensive research on BCA and BIA systems. Attached is an Excel spreadsheet I give away for free.
Body comp - Public spreadsheet.xlsx
I used your app with the [1] Nexpure CF586BLE (E8:6B:EA:82:B3:02) and [2] Hume BodyPod (A0:B7:65:95:11:BA) BIA scales. Both synced and paired easily with my laptop (Windows 11 v25H2). Note that the Nexpure CF586BLE is a rebranded Unique CF597BLE scale, made by Shenzhen Unique Scales Co. Ltd, China. The Unique Health app is made by Shenzhen Lefu Scale Co. Ltd, China.
BMI is an outdated metric and is total BS. I'd encourage you to drop its usage.
I would encourage you to verify your equations and conversions. I reviewed your "BLE Body Composition Scale — Data Retrieval & Decoding Protocol" and found several suspect conversions of the impedance data.
Go here for a Glossary of BIA terms
Go here for BIA Governing Equations
Here are a few comparisons between Blue2Scale, HumeHealth, and Unique Health apps with the Hume and Nexpure scales:
Wt (lbs): B2S Hume = 172.9; B2S Unique = 177.3;
Hume = 172.7; Unique = 177.4
BFM (%): B2S Hume = 27.0; B2S Unique = 28.9;
Hume = 21.2; Unique = 24.0
SMM (lbs): B2S Hume = 79.6; B2S Unique = 79.4;
Hume = 84.2; Unique = 76.2
TBW (%): B2S Hume = 52.0; B2S Unique = 53.3;
Hume = 44.5*; Unique = 55.7
BMR (kcal): B2S Hume = 1582; B2S Unique = 1583;
Hume = 1609; Unique = 1689
VFI: B2S Hume = 12; B2S Unique = 11;
Hume = 8; Unique = 9
SFM (%): B2S Hume = 25.1; B2S Unique = 23.5;
Hume = 18.5; Unique = 20.8
In general, I think your work is brilliant! Since you built this BLE from scratch, I'm guessing that you can add other meaningful BCA metrics that other apps leave out. Namely,
ICW/ECW (cellular health)
ECW/TBW (water balance)
ICW/SLM (muscle quality)
TBW/SLM (muscle hydration)
FFM = Wt - BFM = SLM + (MC + IS) = BCM + ECM
PM = FFM - (TBW + MC + IS)
SLM= TBW + PM
DLM = PM + MC
BCM = ICW + PM
ECM = ECW + MC + IS
LBM = SLM + MC
MA = (10Wt + 6.25Ht + 5 - BMR) / 5
BMR = 370+21.6*SLM / 2.20462
WHR = WC/HC
ASLM = SLM(R-arm) + SLM(L-arm) + SLM(R-leg) + SLM(L-leg)
ASLMI = ASLM/Ht^2
FFMI = FFM/Ht^2
BFMI = BFM/Ht^2
Your impedance package suggests 2 frequencies, 50 and 250 kHz. The 2024 Hume data states the BodyPod 5 and 50 kHz.; whereas, the 2026 Hume data states 20 and 100 kHz. Can you tell from the BE packets the actual impedance frequency? Or are you guessing at low and high?
Given what you know from the impedance data packet, can you calculate the segmental resistance and phase angles, and the whole body phase angle? If so, you can turn Blue2Scale into a very powerful app. The biggest hurdle for app accuracy is validation. Since the hardware is a fixed variable, the software is where you can shine.
Absolutely essential. BIA is an indirect method, so device-specific validation studies (not just manufacturer claims) are what separate hype from reality. High ICCs (often >0.90 for lean mass, fat mass, and % body fat) with DXA or 4-compartment (4C) models are the benchmark for DSMF-BIA devices. Many peer-reviewed papers on 8-electrode multi-frequency systems show excellent agreement for whole-body and segmental measures in diverse populations, though limits of agreement are tighter for lean mass than fat percentage in some cases. Bonus points if validation covers varied groups (age, BMI, ethnicity, clinical conditions like obesity or diabetes). This criterion rightly emphasizes correlation plus low bias—raw numbers like ICC alone aren’t enough; Bland-Altman plots matter for real-world agreement.
This is the practical clincher. Even the best tech fails without low test-retest variability (e.g., <1–2% CV for key metrics under standardized conditions). Clinical-grade DSMF-BIA devices excel here when protocols control for hydration, posture, fasting, exercise, and time of day. High reproducibility makes them useful for tracking changes over time (diet, training, recovery), not just one-off snapshots.
If you are interested in collaborating, I know nothing about programming in BLE, but I have a Master's in engineering and have done extensive research on BCA and BIA systems. Attached is an Excel spreadsheet I give away for free.
Body comp - Public spreadsheet.xlsx