← Elevo
How we measure your posture
Elevo doesn't judge posture by eye. We measure angles used in postural photogrammetry studies, from anatomical points found automatically on your photo by a body-detection model (MediaPipe, Google). You can adjust them, and we show you how certain each value is.
In short: Elevo is a wellness app. It is not a medical device, it doesn't diagnose, and it doesn't replace a check-up. If you have pain, numbness or a spinal condition, we recommend seeing a professional.
1. What each part does
- Body recognition (on your phone). A body-detection model (MediaPipe Pose Landmarker, Google) finds your ear, shoulder and ankle and checks that you're standing with your shoulder toward the phone. From the in-app camera we take 5 quick frames and pick the best one.
- Base of the neck (C7 vertebra). The model doesn't detect it, so we look for it on your silhouette's outline, behind the neck, at the level of its base.
- Your check. We show you the points zoomed in on your head and shoulders. If they're not in the right place, you can move them.
- The angles. We calculate them geometrically from the points, as in clinical photogrammetry.
- Margin of error. It includes the precision of the points, the differences between frames and how upright the phone was (from the phone's sensors; for gallery photos we assume an unknown tilt). If the margin comes out too large, we don't show the value and tell you the photo doesn't allow a precise measurement.
2. What we measure
| Measurement | How it's calculated | Reference used |
|---|
Head position
craniovertebral angle (CVA) | The angle between the horizontal through C7 and the C7–tragus line | ≥ 50° within range; 47–50° borderline; < 47° below range. Thresholds in studies vary (roughly 47–55°). |
Shoulder position
forward shoulder angle (FSA) | The angle between the vertical through C7 and the C7–acromion line | ≥ 52° = forward shoulders (threshold used in studies) |
| Body lean | The angle between the vertical and the ankle bone (malleolus)–C7 line | Adult average 1.73° ± 0.94°; we treat 0–3.6° as usual |
3. How precise it is
- In studies, manual CVA measurement on photographs, with markers placed by a physiotherapist, has good reliability: ICC between 0.85 and 0.94. The minimal detectable change is roughly 2.6–5.9°, depending on the method.
- The margin we show starts from an assumption: automatically found points are accurate to about 1 cm (⅜ in). An unknown phone tilt adds a few degrees.
- For head position, the typical margin is about ±8–9°. That's why we say “clear” when the value is far from the threshold and “probable” when it's close.
- We only measure body lean when your ankles are visible and you use the in-app camera, because it's very sensitive to phone tilt. The camera reads how rotated the phone is and corrects the angles automatically.
- Shoulder position is the least precise of the three, because the C7 and acromion points are close together in a side-on photo.
- On a rescan, we only call a change “real” if it exceeds the measurement error: at least 5° for head and shoulders and 1.5° for lean, or more if the margins of the two scans require it.
4. What we do NOT measure
Spinal curves (kyphosis, lordosis), pelvic tilt and side-to-side asymmetries (such as scoliosis). These need markers on the skin, photos from other angles or a clinical exam. We don't measure height either, and we don't promise inches.
5. The exercises
The plan uses general exercises (activating the deep neck flexors, thoracic mobility, chest stretching, strengthening the muscles between the shoulder blades, glutes, hip flexors). A meta-analysis of 7 randomized trials (627 participants) found that therapeutic exercise can improve CVA in people with a forward head. Individual results vary.
Sources
- Posture Analysis in the Sagittal Plane – Practical Guidelines with Reference Values, Anatomia, 2025, 4(2):5 (CVA, body lean, photo protocol)
- Psychometric Proprieties of a Mobile Application to Measure the Craniovertebral Angle, IJERPH, 2020, 17(18):6521 (smartphone validation, MDC, 50° threshold)
- Reliability of Image J. Software to Evaluate the Craniovertebral Angle in Participants with and without Neck Pain, Archives of Bone and Joint Surgery, 2025, 13(12):849–856 (ICC, SEM, MDC)
- Comparison of Scapulohumeral rhythm in individuals with and without Forward Shoulder Posture, F1000Research, 2024, 13:1421 (FSA definition and the 52° threshold, after Thigpen et al., 2010)
- Effectiveness of Therapeutic Exercise on Forward Head Posture: A Systematic Review and Meta-analysis, Journal of Manipulative and Physiological Therapeutics, 2018, 41(6):530–539 (exercise and CVA)
- Gait analysis comparison between manual marking, 2D pose estimation algorithms, and 3D marker-based system, Frontiers in Rehabilitation Sciences, 2023, 4:1238134, and BMT 2023 (MediaPipe limitations)
Last updated: 22 October 2026.