Kino Vision
Validation How It Works Who It's For Book a demo

Validated against DEXA — body composition from any smartphone.

Objective body fat and lean mass in 30 seconds. No equipment, no calibration, no capital cost.

2.8 % Body fat % MAE vs DXA
0.97 CCC Appendicular lean mass agreement vs DXA
30 sec capture · $377K research funding · n=195 held-out cohort
Kino Vision measurement result

From phone to measurement in 30 seconds.

Guided capture on any smartphone. Structured body composition results in under a minute.

Validation

Strong agreement with DXA across every measure.

Held-out validation against DXA on an independent cohort (n=195). Agreement is strong across body fat and lean mass, and strongest on appendicular lean mass.

Measure
Config
MAE
Lin's CCC (95% CI)
Bland-Altman bias & 95% LoA
Body Fat %
image + inputs
2.77 %
0.930 (0.911–0.943)
+0.98 % [−5.36, +7.33]
Total Lean Mass %
image only
2.85 %
0.916 (0.894–0.933)
−0.60 % [−7.47, +6.28]
Appendicular Lean Mass
image + inputs
3.02 lbs
0.970 (0.962–0.977)
+0.43 lbs [−7.14, +8.00]

Agreement with DXA is strong across all three measures (CCC 0.92–0.97), and strongest on appendicular lean mass. Total lean mass % is estimated from images alone.

Performance in tracking change over time — including lean-mass preservation during GLP-1/incretin-based weight loss — is the focus of ongoing work.

Bland-Altman agreement plot for appendicular lean mass vs DXA

Bland-Altman agreement, appendicular lean mass vs DXA (n=195 held-out cohort)

Scan-to-scan repeatability: pending

Developed on a separate dataset collected independently of the validation study, then evaluated here on an independent held-out cohort (n=195) against DXA.

Study population (n=195): mean age 34.2 (18–75) · 51% male / 49% female · mean BMI 26.8 · DXA body fat 7.5–50.6% (mean 27.8%)

Per-scan confidence estimates: every result includes a reliability estimate, flagging captures where lighting, pose, or framing reduce confidence so they can be retaken.

Validated in IRB-approved studies at two academic medical centers.

Preprint

Reynolds T, Gerard J, Jordan A, Streby N, Stoll B, Bowers A, Hewitt A, Bargamian J, Sapper T, Burdick TE, Kackley M.
"Validation of a Smartphone-Image-Based Computer-Vision Model for Lean Mass and Body Fat Estimation Against Dual-Energy X-ray Absorptiometry."
medRxiv 2026. https://doi.org/10.64898/2026.06.15.26355736

Not yet peer-reviewed.

→ Read the full preprint
Kino Vision provides body composition measurements for informational and research purposes. It is not FDA-cleared and is not intended to diagnose, treat, cure, or prevent any disease. Measurements are not a substitute for professional medical advice.

How it works

01

Capture

Guided photos using any smartphone. Works in clinic, at home, or during virtual visits. Takes 30 seconds.

02

Analyze

Models process the images and estimate body composition. Results ready in seconds.

03

Act

Structured report — body fat %, lean mass, appendicular lean mass, trend over time.

Who it's for

Clinical care

Objective body composition at each visit. No in-clinic scan, no ionizing radiation.

Research

Standardized, remote body composition capture at scale. Consistent methodology, reduced study cost.

Population-scale measurement

Deploy to any smartphone. Aggregate de-identified analytics. Standardized measurement at scale.

Regulatory

Kino Vision is actively pursuing FDA clearance. Current deployments are for research and informational use.

Book a 15-minute walkthrough.

Book a demo