Collecting biomarkers is just the first step. The real value comes from transforming raw data into actionable intelligence that users can actually understand and act on.
The Hard Truth About Raw Data
Users don't care that their HRV was 42ms yesterday and 48ms today. They care that their body is ready for an intense workout or needs active recovery. Intelligence layers make this translation.
The Raw Data Problem
❌ Raw Biomarker Data
- "Your HRV is 42ms"
- "You slept 6.3 hours"
- "Step count: 8,247"
- "RHR: 68 bpm"
User reaction: "So what? Is that good or bad? What should I do?"
Result: Low engagement, high churn
✅ Intelligence Layer Output
- "Recovery score: 68% → Light workout recommended"
- "Sleep quality declining → Review bedtime routine"
- "Activity trend: 15% above baseline → Great week!"
- "Stress increasing → Consider rest day"
User reaction: "I know exactly what to do next."
Result: High engagement, actionable insights
Why Most Apps Fail at Engagement
According to app engagement research, 90% of health apps are abandoned within 30 days. The primary reason? Users get data dumps without context or guidance. They don't know what the numbers mean or what to do about them.
The 4 Intelligence Layer Types
1. Scores (0-3 scale)
Aggregate multiple biomarkers into a single, understandable metric:
- Sleep Score: Overall sleep quality
- Activity Score: Movement patterns
- Readiness Score: Recovery status
- Mental Wellbeing: Mood indicators
- Overall Wellbeing: Holistic health
Example: "Sleep Score: 2.1/3 (Good) - Your sleep duration and quality are both strong."
2. Archetypes (14 classifications)
Classify user behavior patterns for personalized recommendations:
- Sleep Archetypes: "Consistent Early Sleeper", "Irregular Sleeper"
- Activity Archetypes: "Highly Active", "Moderately Active", "Sedentary"
- Wellness States: "Thriving", "Stable", "Struggling"
Example: "You're a 'Weekend Warrior' - high weekend activity, low weekday movement."
3. Trends (7-day, 14-day, 30-day)
Identify patterns and predict future behavior:
- Improving: Metrics trending upward
- Declining: Early warning signals
- Stable: Consistent patterns
- Volatile: High variability
Example: "Your sleep quality has declined 18% over 14 days - stress levels may be impacting rest."
4. Comparisons (Peer benchmarks)
Contextualize data against similar users:
- Age group: vs. 30-39 year olds
- Activity level: vs. similar fitness
- Demographics: vs. similar cohort
Example: "Your activity is in the top 25% for your age group."
Real-World ROI Examples
Fitness App: Engagement Metrics
Case Study: Fitness App with Intelligence Layers
Before: Raw step counts, heart rate graphs → 68% monthly churn
After: Recovery scores, workout readiness, trend alerts → 32% monthly churn
Result: 2.1x increase in LTV, 3.4x longer session durations
Insurance Program: Risk Detection
Why This Matters for Insurance
Intelligence layers detect mental health deterioration 3-5 days before users self-report symptoms. This allows proactive interventions (EAP programs, telehealth, resources) before crises occur, reducing ER visits and inpatient care costs.
Mental Health Platform: Clinical Outcomes
The Cost of Building vs. Buying Intelligence
| Component | Build In-House | Buy (Sahha) |
|---|---|---|
| Data Science Team | $400k-$800k/year (3-5 engineers) | Included |
| Algorithm Development | 12-18 months minimum | Immediate access |
| Clinical Validation | $200k-$500k (research studies) | Pre-validated (18+ published studies) |
| Ongoing Maintenance | $150k-$300k/year | Included in subscription |
| Model Updates | Manual (quarterly at best) | Continuous (ML pipeline) |
| Total First Year Cost | $750k - $1.6M | ~$50k (typical deployment) |
⚠️ The Hidden Costs of Building
Beyond direct costs, building intelligence layers in-house requires:
- Domain expertise: Sleep science, behavioral psychology, clinical validation
- Data infrastructure: ML pipelines, model training, A/B testing frameworks
- Regulatory compliance: HIPAA, GDPR, medical device regulations (if clinical)
- Opportunity cost: Engineering resources diverted from core product
Most companies underestimate build time by 3-4x and total cost by 5-10x.
Platform Comparison: Who Offers Intelligence?
| Platform | Intelligence Layers | Clinical Validation | Use Case Fit |
|---|---|---|---|
| Sahha | 5 Scores 14 Archetypes Trends MCP Integration | 18+ published studies | Consumer apps, insurance, mental health, AI agents |
| Terra | Raw data only | None (data aggregator) | Build your own intelligence |
| Rook | Raw data only | None (data aggregator) | Build your own intelligence |
| Spike | GenAI processing | Medical equipment focus | Clinical trials, EMR integration |
| Vital/Junction | Lab data only | Lab test validation | Lab integration only (wearables sunset) |
Why Only Sahha Offers Consumer-Ready Intelligence
Terra and Rook are data aggregators - they unify wearable APIs but provide zero intelligence. You get raw step counts, heart rate, sleep duration, etc. You still need to build:
- Scoring algorithms
- Behavioral classification models
- Trend detection systems
- Clinical validation studies
Sahha is the only platform that delivers validated, production-ready intelligence layers out of the box.
The MCP Advantage: Intelligence for AI Agents
Sahha's Model Context Protocol (MCP) integration enables AI agents to access pre-processed intelligence layers directly:
Raw Data for AI Agents
Problem: AI must process raw biomarkers every query
- High token usage (passing full datasets)
- Inconsistent interpretations
- No behavioral context
- Slow response times
Intelligence Layers for AI Agents
Solution: AI gets pre-processed insights via MCP
- Low token usage (scores + context)
- Validated interpretations
- Rich behavioral profiles
- Instant responses
Example MCP integration:
Agent: "How is the user's wellness today?"
MCP Response: {
"sleep_score": 2.1,
"state": "good",
"archetype": "Consistent Early Sleeper",
"trend": "stable_7d",
"mental_wellbeing": 1.8,
"recommendation": "Maintain current sleep routine, readiness is high"
}
The AI agent receives actionable intelligence instead of raw numbers, enabling better coaching, personalized responses, and predictive insights.
When Intelligence Layers Are Essential
You NEED intelligence layers if your app/platform:
- ✅ Provides personalized recommendations (workout plans, nutrition, mental health)
- ✅ Requires behavioral predictions (churn risk, health deterioration, adherence)
- ✅ Drives user engagement (gamification, streaks, progress tracking)
- ✅ Supports clinical decisions (insurance risk, care escalation, treatment plans)
- ✅ Powers AI agents (health coaches, chatbots, virtual assistants)
You DON'T need intelligence layers if you:
- Only display passive dashboards (no recommendations)
- Export data for external analysis (BI tools, data warehouses)
- Build entirely custom ML models (research-focused)
In these cases, raw data aggregators (Terra, Rook) may be sufficient.
Getting Started with Intelligence Layers
Sahha's intelligence layers are available immediately via:
- REST API: Query scores, archetypes, trends on-demand
- Webhooks: Receive real-time intelligence updates
- MCP Integration: AI agents access via Model Context Protocol
- Batch Processing: Analyze cohorts for research/analytics
Next Steps
- 📖 Intelligence Layers Breakdown - Detailed documentation of all scores, archetypes, trends
- 📱 Smartphone vs Wearable Data - Collection methods and trade-offs
- 🔧 Integration Approaches - How to access intelligence layers
- 🎯 Choosing a Health Data API - Platform comparison guide