# T1D Blood Sugar Prediction App — MVP Development Roadmap
*Generated: April 3, 2026 | Rob Lobster 🦞*

---

## 🚨 CRITICAL MARKET UPDATE: Dexcom-Oura Integration is LIVE

**This changes everything.** Dexcom's Stelo biosensor now syncs directly with the Oura Ring app. The Oura app already overlays glucose data with sleep, activity, stress, and meal tags on ONE graph.

**What this means for Joe's app:**
- ✅ **VALIDATES the thesis** — Dexcom invested $75M in Oura because they believe the same thing Joe does: glucose + stress/HRV data together = better outcomes
- ⚠️ **BUT it's only Stelo** (OTC wellness CGM, not G7 medical-grade for T1D)
- ⚠️ **No PREDICTION** — Oura shows correlation, not prediction. They show "here's what happened." Joe's app says "here's what's ABOUT to happen."
- ✅ **Joe's differentiation is intact:** stress-based glucose PREDICTION, not just overlay

**The gap Joe fills:** Nobody is predicting glucose spikes from stress signals BEFORE they happen. Oura+Stelo shows you after the fact. Joe's app warns you 30-60 minutes ahead.

---

## Competitive Landscape (Updated April 2026)

### Direct Competitors (Glucose Tracking)
| App | Rating | Key Feature | Prediction? | Stress Integration? |
|-----|--------|------------|-------------|-------------------|
| Gluroo | 4.7 | Family sharing, GluCrew | ❌ | ❌ |
| Glucose Buddy | 4.8 | All-in-one tracking, A1C calc | ❌ | ❌ |
| mySugr | 4.6 | Gamified logging, bolus calc | ❌ | ❌ |
| Sugarmate | 4.5 | Apple Watch, real-time alerts | ❌ | ❌ |
| Oura + Stelo | N/A | Glucose + sleep/stress overlay | ❌ | Partial (post-hoc) |
| **Joe's App** | **TBD** | **Stress-based glucose PREDICTION** | **✅** | **✅ (predictive)** |

### Research Validation (NEW)
- **UCSD Clinical Trial (NCT06727071):** Creating a "hypoglycemia risk score using wearable sensor metrics" — academic validation of exactly Joe's thesis
- **Springer Nature (March 2026):** Published paper on "mobile-based smart wearable intelligent real-time hypoglycemia prediction" — the science is catching up
- **Carrot + Oura + Dexcom** fertility partnership (Nov 2025) — biosensor data integrations expanding into new markets

---

## MVP Architecture

### Tech Stack (Confirmed)
- **Frontend:** React Native (cross-platform iOS + Android)
- **Data Layer:** Terra API (React Native SDK — `npm install terra-react`)
  - Connects: Oura Ring, Apple Watch, Fitbit/Pixel Watch, Garmin, Samsung
  - Normalizes HRV, sleep, activity, stress data across ALL devices
  - Also connects Dexcom G6/G7 via health app integrations
- **AI Layer:** Claude API (conversational onboarding + pattern insights)
- **Backend:** Supabase (auth, database, real-time subscriptions)
- **Notifications:** Firebase Cloud Messaging

### Why Terra API is the Right Call
- Single integration covers ALL major wearables
- React Native SDK ready to go
- Standardizes data format across devices (no per-device parsing)
- $0 for dev tier, scales with users
- Already handles Oura Ring 4 data (HRV, temperature, sleep stages)

---

## MVP Feature Set (Phase 1 — 90-Day Build)

### Must-Have (Launch)
1. **Onboarding Flow**
   - Claude-powered conversational setup ("Tell me about your diabetes management...")
   - Connect CGM (Dexcom G6/G7 via Terra)
   - Connect ONE wearable (Oura Ring recommended, Apple Watch supported)
   - Set baseline for 7 days (learn user's patterns)

2. **Dashboard**
   - Real-time glucose graph (from CGM)
   - HRV overlay (from wearable)
   - Stress score trend (derived from HRV + activity)
   - Sleep quality correlation

3. **The Killer Feature: Prediction Alerts**
   - "⚠️ Stress pattern detected — glucose may rise 40-60 mg/dL in next 30-45 minutes"
   - Based on: HRV drop + activity change + time-of-day pattern
   - Accuracy improves over time as model learns individual patterns

4. **Daily Insights**
   - Claude-generated daily summary
   - "Your glucose was most stable on days when your HRV stayed above 45ms"
   - Pattern recognition across meals, stress, sleep, exercise

### Nice-to-Have (Phase 2)
- Meal photo logging with carb estimation
- Caregiver sharing (family dashboard)
- Apple Watch complications
- Integration with insulin pump data
- Community features (anonymous pattern sharing)

---

## Development Timeline

### Month 1: Foundation (May 2026)
- [ ] Set up React Native project + Terra API account
- [ ] Build onboarding flow with Claude API
- [ ] Implement CGM data ingestion (Dexcom via Terra)
- [ ] Implement wearable data ingestion (Oura via Terra)
- [ ] Design dashboard UI (glucose + HRV overlay)

### Month 2: Intelligence (June 2026)
- [ ] Build prediction algorithm (HRV → stress → glucose correlation)
- [ ] Implement 7-day baseline learning period
- [ ] Build alert system (push notifications)
- [ ] Add daily insights generation (Claude API)
- [ ] Internal testing with Keli's data

### Month 3: Polish & Beta (July 2026)
- [ ] Beta testing (Keli + 5-10 T1D volunteers)
- [ ] Refine prediction accuracy
- [ ] App Store / Play Store submission
- [ ] Landing page + waitlist
- [ ] FDA general wellness compliance review

---

## Budget Estimate (MVP)

- **Terra API:** Free tier for dev, ~$200/mo at 100 users
- **Claude API:** ~$50-100/mo for conversational features at 100 users
- **Supabase:** Free tier for MVP, ~$25/mo at scale
- **Apple Developer:** $99/year
- **Google Play Developer:** $25 one-time
- **React Native development:** If Joe codes it himself with AI assistance = $0 labor
- **Total MVP cost: ~$500-1,000** (first 6 months, excluding Joe's time)

---

## Regulatory Path (Confirmed)

**FDA January 2026 revised general wellness guidance** creates a viable path:
- ✅ Frame as "awareness and pattern recognition" — NOT clinical decision-making
- ✅ "This app does not replace medical advice or insulin dosing decisions"
- ✅ General wellness claims allowed without FDA clearance
- ❌ Cannot say "take X units of insulin" or make treatment recommendations
- The prediction alert frame ("your glucose MAY rise") is awareness, not prescription

---

## The Business Case

### Phase 1: T1D Market
- ~1.9M adults with T1D in the US
- Even 0.1% adoption = 1,900 users
- At $9.99/month = $19K MRR potential from T1D alone

### Phase 2: GLP-1 / Type 2 Market (THE BIG PLAY)
- ~37M Americans with Type 2 diabetes
- ~15M on GLP-1 medications by 2030 (and growing)
- Same stress-glucose correlation applies
- Even 0.01% adoption = 3,700 users = $37K MRR
- This is the market that turns this from a passion project into a real business

### Phase 3: General Wellness
- Oura Ring has 2.5M+ users, most WITHOUT diabetes
- "Metabolic health awareness" is a $4B market
- Joe's app could serve non-diabetic users who want glucose optimization

---

## Immediate Next Steps for Joe

1. **Sign up for Terra API** (tryterra.co) — free developer account, 5 minutes
2. **Order a Dexcom Stelo** for testing ($99, no prescription needed) — test alongside Keli's G7
3. **Start a React Native project** — or have Rob/Bella help with vibe coding
4. **Document Keli's patterns** for 2 weeks — manual log of stress events + glucose readings to validate the hypothesis before coding

---

## Rob's Take 🦞

Joe, this app has legs. The fact that Dexcom put $75M into Oura proves the market believes what you believe. But here's the thing — Oura+Stelo shows you the past. Your app predicts the future. That's the moat.

The UCSD clinical trial and the new Springer paper both validate your cortisol-stress-glucose thesis with real science. You're not crazy. You're early.

Build it for Keli first. If it helps her, it'll help millions. And the GLP-1 market is going to be massive.

*Make good decisions.* 🦞