Zyntra

• 

  Predicted 1 hour in advance

• 

  Actionable guidance in real time

• 

  A complete prediction-to-action flow

• 

  Arquitecture

Inspiration

Healthcare doesn’t fail because of treatment. It fails because of timing.

In chronic conditions like diabetes, patients and clinicians often act when symptoms appear — when it’s already too late.

We asked ourselves: What if we could anticipate risk before the body shows symptoms?

---

What it does

Zyntra is a predictive intelligence layer for healthcare that anticipates critical events before they happen.

Starting with diabetes, it analyzes real-time data from continuous glucose monitors (CGMs), wearables, and behavioral signals to predict glucose events 1–24 hours in advance.

More importantly, Zyntra doesn’t just predict — it provides clear, actionable recommendations so patients and clinicians can act early.

---

How we built it

Zyntra is powered by a multi-agent architecture:

• Data Agent → processes real-time health signals
  
• Risk Agent → predicts future events using AI
  
• Coach Agent → generates personalized recommendations
  

We implemented a hybrid predictive model:

• A general model trained on population data
  
• A personalized model that adapts after ~30 days of user data
  

This allows Zyntra to evolve from generic insights to highly personalized predictions.

---

Challenges we ran into

• Limited access to high-quality real-world healthcare data
  
• Ensuring predictions are interpretable and trustworthy
  
• Translating complex outputs into simple, actionable guidance
  

---

Accomplishments that we're proud of

• Built a working prototype with real-time data simulation
  
• Achieved ~75% prediction reliability on the available dataset
  
• Designed a fully functional multi-agent system
  
• Created a clear prediction → action flow for users
  

---

What we learned

• Prediction without action has no value
  
• Simplicity is critical in healthcare communication
  
• Trust and explainability are as important as accuracy
  

---

What's next for Zyntra

• Clinical validation with real patient data
  
• Integration with hospital systems and CGM providers
  
• Regulatory pathway (CE Mark as SaMD)
  
• Expansion beyond diabetes into other chronic conditions
  

---

Zyntra is not another monitoring tool.

It’s a shift from reactive care to predictive care.

We don’t help doctors react faster.
We help them act before the problem exists.

Built With

• csv-data-processing
• decision
• llm-based
• machine-learning-(hybrid-predictive-models)
• multi-agent-architecture
• next.js
• node.js
• python
• react
• rest-apis
• typescript
• vercel