All photos and scan results are encrypted in transit and at rest. Identifiable features can be minimized, and only you can access your history. You are always in control of your data.
You can permanently delete your scans and associated data from within the app or via our Data Deletion page. When you delete it, it’s gone.
Body weight and BMI are crude indicators that do not distinguish between fat and muscle. Two people with the same BMI can have very different physiques and health risk profiles, depending on their body fat percentage and distribution. For tracking fitness and physique changes, body fat % is a more informative metric than weight alone. Gold-standard tools like dual-energy X-ray absorptiometry (DEXA) quantify fat and lean mass directly, but they are not designed for frequent, everyday use. BodyFat AI exists to bring DEXA-style insight into a practical, app-based format that can be used regularly.
Dual-Energy X-ray Absorptiometry (DEXA) is widely considered the reference method for body composition assessment. DEXA provides precise estimates of total and regional fat mass, lean mass, and bone mineral content, and is often used as the comparator in validation studies of new methods [1]. However, DEXA scanners are expensive, require trained staff, and are typically located in hospitals, imaging centers, or research facilities [1]. Each scan involves a clinic visit, takes time to schedule, and includes a small but non-zero dose of X-ray exposure. As a result, DEXA is excellent for occasional benchmarking but impractical for weekly or monthly tracking in the general population.
Bioelectrical Impedance Analysis (BIA) is the most common consumer alternative to DEXA. Home body fat scales and gym-based BIA scanners (including multi-frequency and segmental devices) estimate body composition by passing a weak electrical current through the body and inferring fat mass from resistance. BIA’s strengths are convenience and speed, but its accuracy is highly sensitive to hydration and other factors. Studies show that changes in fluid balance, recent food or drink intake, exercise, and skin temperature can meaningfully affect impedance, and thus the body fat estimate [2,3]. Even under controlled conditions, BIA tends to show only moderate agreement with DEXA. In a large adult cohort, multi-frequency BIA had a concordance correlation around 0.9 with DEXA for body fat percentage—reasonable but clearly below the near-perfect agreement expected of a gold standard [2]. Errors were larger in individuals with overweight or obesity [2], which limits reliability in precisely those users who often most need accurate tracking.
Consumer smart scales show even greater variability. An observational study comparing several commercially available scales with DEXA found substantial differences in fat mass and concluded that such devices are not sufficiently accurate to replace DEXA for body composition assessment [4]. In practice, this means BIA-based devices can be useful for rough trends but may be off by several percentage points in body fat, and day-to-day fluctuations may reflect hydration more than real tissue change.
Skinfold calipers estimate body fat by measuring the thickness of pinched skinfolds at multiple sites and applying equations to approximate total body fat percentage. They are inexpensive and, in skilled hands, can track changes in subcutaneous fat reasonably well. However, calipers are limited by operator dependence and assumptions. Accurate readings require a trained technician to locate sites and apply consistent pressure. They primarily measure subcutaneous fat, assuming a typical relationship between subcutaneous and visceral fat that may not hold for all individuals, especially in obesity or specific medical conditions [5]. When directly compared with DEXA, skinfold-based estimates often underestimate body fat, sometimes by 7–9 percentage points in specific populations [5]. For self-administered or casual use, these limitations make calipers less practical and less reliable than they appear on paper.
Recent advances in computer vision and deep learning have enabled a third path: estimating body fat percentage from standard photographs. Instead of electrical currents or localized pinches, these models analyze body shape, contours, and proportions visible in 2D images, then infer overall body composition. This is the foundation for BodyFat AI.
Several peer-reviewed studies published since 2022 have evaluated AI-based 2D or smartphone photo methods directly against DEXA. A large 2025 study in NPJ Digital Medicine evaluated an AI 2D-photo method in over 1,200 adults. The photo-based estimates of body fat percentage showed very high agreement with DEXA, with a concordance correlation coefficient reported near 0.98 and small average bias [1]. In the same dataset, common field methods such as multi-frequency BIA and skinfolds showed noticeably lower agreement [1]. The authors concluded that AI 2D-photo assessment can be functionally interchangeable with DEXA for estimating body fat % at the individual level.
Another validation study compared smartphone image-based estimates and a consumer BIA scale against DEXA. The photograph-based method showed stronger agreement with DEXA than the impedance scale, indicating that shape-based AI can outperform devices that rely on electrical signals for estimating body fat [6]. Earlier work has similarly shown that deep learning systems trained on large datasets of people with known body composition can infer body fat percentage from a small number of standardized photos with mean absolute errors of only a few percentage points relative to DEXA [6]. Across studies, the pattern is consistent: AI methods using well-captured images can match or exceed the accuracy of other common at-home tools when compared to DEXA.
BodyFat AI uses this same class of approach: standardized multi-pose images plus basic data (such as height and weight) feed into an AI model refined on large datasets. Because the model has learned complex relationships between visible body shape and underlying composition, it can provide a single body fat % estimate that closely tracks what a DEXA scanner would report, while requiring only a phone camera.
Putting all of this together:
Used regularly and under consistent conditions, BodyFat AI provides a realistic, research-aligned estimate of your body fat percentage and a clear view of how it changes over time. For most people, it offers a closer and more practical alternative to DEXA than traditional consumer methods, while remaining simple, private, and affordable.
Disclaimer: BodyFat AI is intended for general wellness and fitness purposes only and is not a medical device. It should not be used to diagnose, treat, cure, or prevent any disease. Always consult a qualified healthcare professional for medical advice or concerns about your health.
