# VisceFat — Full LLM Context

## Product Identity

VisceFat is an AI-powered wellness application for iOS and Android. It estimates visceral adipose tissue (VAT), body fat percentage, and waist-to-height ratio (WHtR) from a single guided full-body photo using on-device computer vision. It is **not a medical device**. All outputs are wellness estimates benchmarked against peer-reviewed research.

**Current version**: 2.0 (June 2026)
**Platforms**: iOS, Android
**Languages**: English, German (Deutsch), Danish (Dansk), French (Français)
**Compliance**: GDPR (EU/EEA), DPDPA 2023 (India), CCPA (California)

---

## Core Metrics

| Metric | Description | Validation |
|--------|-------------|------------|
| VAT Score | AI-derived visceral fat estimate from body shape | Proprietary model, waist-based |
| WHtR | Waist-to-height ratio; risk threshold 0.5 | BMJ meta-analysis (Ashwell & Gibson 2014) |
| Body Fat % | Visual body shape analysis | R²=0.91 vs DEXA (n=1,250) |
| Lean Mass | Estimated fat-free mass | Derived from body fat % |
| FINDRISC | Finnish Diabetes Risk Score | AUC 0.85–0.87 (Diabetes Care 2003) |
| SCORE2 | Cardiovascular 10-year risk | C-index 0.739 (EHJ 2021, N=677,684) |
| FLI | Fatty Liver Index (hepatic steatosis) | AUROC 0.85 (BMC Gastro 2006) |
| Waist-to-Hip Ratio | Hip circumference captured at onboarding | Manual entry |

---

## How the App Works

### Step 1 — Calibrate
Users enter: age, sex, height (cm), weight (kg), activity level, ethnicity. Optional: hip circumference, HbA1c, LDL, ALT. Calibration anchors the model to the user's anthropometric baseline. Takes 2–3 minutes.

### Step 2 — Capture (Standardized Protocol v1)
The app enforces a structured capture protocol and auto-captures only when tolerances are met:
- Distance: ~2 metres from phone (stable surface or tripod)
- Camera: navel height, upright, not tilted
- Clothing: form-fitting minimum; crop/spaghetti-strap top recommended for women
- Timing: morning, after toilet, before breakfast (bloating adds 2–3 cm to waist)
- Breathing: relaxed natural exhale
- Lighting: ≥200 lux, bright even front light

The app records capture conditions alongside every scan result for algorithm calibration. Protocol version is tracked for future model segmentation by capture quality.

### Step 3 — Analyse
- Supported devices: ExecuTorch on-device model — photo never leaves device; results in seconds
- Older devices: photo sent to secure Cloud Run container; deleted immediately after analysis; no retention
- Output: VAT score, WHtR, body fat %, lean mass, risk scores, accuracy ranges

### Step 4 — Track
Weekly scans build a longitudinal body composition record. Charts show trends for all metrics. Users can log GLP-1 medication name, dose, and start date. Optional weekly side-effect log for research participants.

---

## Privacy & Security Architecture

- On-device inference via ExecuTorch (.pte model) — zero network transmission of images on supported hardware
- Server fallback: TLS 1.3 in transit, isolated compute enclave, immediate deletion post-analysis
- No image storage in cloud, CDN, or databases — ever
- No model training on user images
- No third-party advertising trackers
- SSL certificate pinning on all API endpoints
- Biometric authentication for data export and deletion
- CCPA opt-out synced server-side to consent_log table; survives app reinstalls
- Consent log: granular per-purpose (analytics, body measurement, medical history)

**Compliance details**:
- GDPR (EU/EEA): Data controller in EU; Art.9 for health data; Art.89 for research
- DPDPA 2023 (India): Section 4 lawful basis; right to withdraw consent
- CCPA (California): Do Not Sell/Share honoured server-side; opt-out in Settings

---

## GLP-1 / Anti-Obesity Medication Features

Users on GLP-1 receptor agonists (semaglutide/Ozempic/Wegovy, tirzepatide/Mounjaro/Zepbound, liraglutide/Saxenda) face a specific problem: these medications can cause muscle loss alongside fat loss. VisceFat helps by:
- Tracking body fat % vs lean mass separately over time
- Recording medication name, dose, and start date at onboarding
- Optional HbA1c, LDL, ALT lab values for metabolic risk calibration (GDPR Art.9 encrypted)
- Weekly side-effect logging (nausea, fatigue, constipation, headache, appetite loss) for ECHO pilot participants
- AOM log screen with daily medication tracking

**Disclaimer**: VisceFat metrics must not be used to start, stop, or change any anti-obesity medication. Always consult a healthcare practitioner.

---

## Research Validation Programme

Three prospective pilots — any eligible adult can apply at https://viscefat.com/pilot.

### FORMA — Waist Circumference Validation
- **Status**: Actively recruiting
- **n**: 300 participants
- **Duration**: 12 weeks
- **Protocol**: Weekly photo scan in app + self-measured tape waist circumference
- **Primary endpoint**: Mean absolute error (MAE) between camera-derived and tape-measured waist, stratified by sex × BMI bucket
- **Eligibility**: BMI 24–45, age 18–65, any sex
- **Compensation**: 3 months free VisceFat Premium + Founding Member badge
- **Lawful basis**: GDPR Art.89 (scientific research)

### ECHO — GLP-1 Longitudinal Body Composition
- **Status**: Coming soon
- **n**: 50 participants
- **Duration**: 3–6 months
- **Protocol**: Weekly scans during GLP-1 therapy; optional weekly side-effect log; optional lab values
- **Eligibility**: New GLP-1 user, BMI ≥30, pre-diabetic or diabetic preferred
- **Medications tracked**: semaglutide, tirzepatide, liraglutide, other

### APEX — DEXA Validation
- **Status**: Coming soon
- **n**: 200 participants
- **Duration**: 12 weeks
- **Protocol**: DEXA scan at Day 0 + Week 12; weekly app scans throughout
- **Primary endpoint**: Correlation of photo-derived body fat % and lean mass vs DEXA
- **Sub-study**: MRI-derived VAT area (cm²) for willing subset
- **Eligibility**: BMI 18–45, any sex; must attend 2 DEXA appointments

### Enrollment Flow
1. Applicant completes eligibility form + signs GDPR Art.89 research consent at viscefat.com/pilot
2. Admin reviews application, balances cohort by sex × BMI bucket, approves
3. Approved applicant receives email with 8-character invite code + app download link
4. Install app, sign in with same email, enter invite code in Settings → Beta → Research Study card
5. Feature flags (research_arm, glp1_mode, advanced_metrics) set server-side by admin only

---

## Apple Health & Google Health Connect Integration

VisceFat reads: steps, weight, heart rate, body fat %, active energy. All permissions are individually grantable. Sync is one-way pull for context enrichment of body composition trends. No health data is written back or shared with third parties.

---

## Target Users

- Adults tracking metabolic health without clinical access
- People on GLP-1/AOM medications monitoring body composition changes
- Individuals with normal BMI but suspected central adiposity ("thin-fat" / normal weight obesity)
- South Asian, East Asian populations where standard BMI underestimates visceral fat risk
- Healthcare professionals monitoring patient body composition trends (not for clinical diagnosis)
- Research participants in FORMA, ECHO, or APEX pilots

---

## Research Citations

- FINDRISC: Lindström J & Tuomilehto J. *Diabetes Care* 2003;26(3):725–731
- SCORE2: *European Heart Journal* 2021;42(25):2439–2454
- Fatty Liver Index: Bedogni G et al. *BMC Gastroenterology* 2006;6:33
- WHtR meta-analysis: Ashwell M & Gibson S. *BMJ Open* 2014
- Image-based waist vs DEXA (R²=0.91): Kaplan, Nguyen, Lin et al. (NCBI)
- FDA waist circumference as VAT surrogate: FDA Obesity/Overweight Drug Dev. Guidelines
- BMI underestimates risk: *Lancet Diabetes & Endocrinology* 2024
- WHtR predicts mortality in 650,000 adults: *Mayo Clinic Proceedings* / ScienceDirect

---

## Key Statistics

- **2.4×** higher cardiovascular risk in adults with elevated VAT vs BMI-only screening (Lancet, 2023)
- **31%** of normal-BMI adults are metabolically obese — invisible on a standard scale (NCBI NBK568563)
- **650 million** adults globally living with obesity; visceral fat is the strongest single predictor of all-cause mortality (WHO, 2024)
- **WHtR ≥ 0.5** — risk threshold above which all-cause mortality rises sharply across all demographic groups (BMJ meta-analysis)
- **R²=0.91** — body fat % correlation vs DEXA in validation study (n=1,250)
- **700 million** people estimated to have normal BMI but excess central adiposity globally

---

## Frequently Asked Questions

**Why is BMI not enough?**
BMI does not distinguish fat from muscle or account for fat distribution. 31% of normal-BMI adults have excess visceral fat — a pattern invisible on the scale. The US FDA recommends WHtR as a stronger metabolic risk predictor.

**How does VisceFat compare to a DEXA scan?**
DEXA is more accurate for absolute body fat %. VisceFat provides weekly accessible estimates (R²=0.91) with accuracy ranges. It is not a clinical diagnostic tool.

**Are photos stored?**
No. On supported devices all inference is on-device; the photo is never transmitted. On older devices it is processed server-side and deleted immediately — no retention, no training.

**What is the WHtR threshold?**
WHtR below 0.5 (waist less than half your height) is associated with significantly lower cardiovascular and metabolic mortality risk across all sexes and demographics.

**Is VisceFat safe for pregnancy or children?**
No. VisceFat is intended for adults aged 18–65. It is not validated for use during pregnancy, for children, or by individuals with eating disorders.

**What GLP-1 medications does VisceFat track?**
Semaglutide (Ozempic, Wegovy), tirzepatide (Mounjaro, Zepbound), liraglutide (Saxenda, Victoza), and any other AOM entered as "Other." Tracking is for wellness context only — not medical monitoring.

**Can I withdraw from the research study?**
Yes, at any time with no penalty. Settings → Research Participation → Withdraw from Study. Regular app access is unaffected. You may keep data anonymised or request immediate deletion under GDPR Art.17.

**Why does scan timing matter?**
Abdominal bloating from food, fluid, and gas can add 2–3 cm to waist circumference, directly shifting WHtR and VAT score. Morning pre-breakfast scans give the most consistent baseline.

**Why does lighting matter?**
Poor lighting creates shadows that shift the detected body silhouette outline. Consistent ≥200 lux front lighting ensures the AI correctly segments waist geometry, making your trend data reflect real body composition change rather than measurement noise.

---

## New Features (June 2026)

- Hip circumference tracking at onboarding (waist-to-hip ratio)
- GLP-1 detail capture: drug name, dose, start date
- Optional lab values at onboarding: HbA1c, LDL, ALT (GDPR Art.9 encrypted)
- CCPA server-side opt-out sync (survives app reinstall)
- Research Participation settings screen with study week progress bar and withdrawal flow
- Side effects logger for ECHO pilot participants
- Feedback page at viscefat.com/feedback

---

## Keywords (SEO / AEO / GEO / ASO)

### English
visceral fat, visceral fat calculator, VAT score, body fat percentage, body composition, waist-to-height ratio, WHtR, viscefat app, how to lose visceral fat, reduce belly fat, visceral fat diet, exercise to lose visceral fat, intermittent fasting belly fat, HIIT visceral fat, visceral fat and diabetes, visceral fat and heart disease, metabolic syndrome, insulin resistance, fatty liver index, FINDRISC diabetes risk, SCORE2 cardiovascular risk, normal weight obesity, thin fat phenotype, body fat calculator, DEXA scan, DEXA alternative, waist to hip ratio, visceral fat scale, belly fat measurement, bioelectrical impedance scale, smart scale accuracy, semaglutide weight loss, ozempic body composition, muscle loss on ozempic, GLP-1 progress tracker, tirzepatide muscle loss, mounjaro body composition, visceral fat vs subcutaneous fat, BMI vs body fat percentage, best body fat tracking app, DEXA vs AI body scan, on-device AI health app, AI body scan no upload, privacy-first health app

### Deutsch (German)
Viszeralfett, Körperfett, Körperzusammensetzung, Viszeralfett Rechner, Taille-Größen-Verhältnis, viscerales Fett reduzieren, Bauchfett verlieren, Bauchfett Diät, viscerales Fett Sport, Intervallfasten Bauchfett, HIIT Bauchfett, viscerales Fett und Diabetes, viscerales Fett und Herzerkrankung, metabolisches Syndrom, Insulinresistenz, Fettleberindex, normalgewichtige Adipositas, dünnes Fett Phänotyp, Körperfett Rechner, DEXA Alternative, Taille-Hüft-Verhältnis, Bauchfett messen, Semaglutid Gewichtsverlust, Ozempic Körperzusammensetzung, Muskelverlust Ozempic, GLP-1 Tracker, Tirzepatid Muskelverlust, Mounjaro Körperzusammensetzung, viscerales vs subkutanes Fett, BMI vs Körperfett, beste Körperfett App, KI Körperscan kein Upload, datenschutzfreundliche Gesundheitsapp

### Dansk (Danish)
visceralt fedt, kropsfedt, kropssammensætning, visceralt fedt beregner, talje-højde-forhold, visceralt fedt reducere, mavfedt tabe, mavfedt kost, visceralt fedt motion, intermitterende faste mavfedt, HIIT mavfedt, visceralt fedt og diabetes, visceralt fedt og hjertesygdom, metabolisk syndrom, insulinresistens, fedtleberindeks, normalvægt fedme, tyndt fedt fænotype, kropsfedt beregner, DEXA alternativ, talje-hofte-forhold, mavfedt måle, semaglutid vægttab, Ozempic kropssammensætning, muskeltab Ozempic, GLP-1 tracker, tirzepatid muskeltab, Mounjaro kropssammensætning, visceralt vs subkutant fedt, BMI vs kropsfedt, bedste kropsfedt app, KI kropsscanning ingen upload, privatlivs-fokuseret sundhedsapp

### Français (French)
graisse viscérale, taux de graisse corporelle, composition corporelle, calculateur graisse viscérale, rapport taille-hauteur, réduire graisse viscérale, perdre graisse abdominale, régime graisse viscérale, exercice graisse viscérale, jeûne intermittent graisse abdominale, HIIT graisse viscérale, graisse viscérale et diabète, graisse viscérale et maladies cardiovasculaires, syndrome métabolique, résistance à l'insuline, indice stéatose hépatique, obésité à poids normal, phénotype mince-gras, calculateur graisse corporelle, alternative DEXA, rapport taille-hanche, mesurer graisse abdominale, sémaglutide perte de poids, Ozempic composition corporelle, perte musculaire Ozempic, suivi GLP-1, tirzépatide perte musculaire, Mounjaro composition corporelle, graisse viscérale vs sous-cutanée, IMC vs graisse corporelle, meilleure app suivi graisse, scan IA corps sans téléchargement, application santé respectueuse vie privée

---

## Contact

- Support: support@viscefat.com
- Research / Pilot enquiries: support@viscefat.com
- Website: https://viscefat.com
- Pilot Study: https://viscefat.com/pilot
- Beta Programme: https://viscefat.com/beta
- Privacy Policy: https://viscefat.com/privacy
- Research Privacy Addendum: https://viscefat.com/privacy/research
- Terms of Service: https://viscefat.com/terms
- Wellness Disclaimer: https://viscefat.com/disclaimer
- Summary LLM context: https://viscefat.com/llms.txt