Design file of our project

Inspiration

India is known as the diabetes capital of the world, with millions suffering from diabetes and other chronic diseases without affordable solutions. As a Type 1 diabetic myself, I have faced the challenges of managing my condition effectively while keeping costs low. Many people struggle with early detection, disease management, and metabolic health optimization, which inspired us to build D-Twina—a revolutionary AI-powered metabolic twin that provides personalized health insights, disease prediction, and biohacking solutions at a fraction of the cost.

What It Does

D-Twina analyzes and predicts an individual’s metabolic health using AI and ML by integrating data from:

Smartwatches & CGMs: Collects real-time health data such as stress levels, sleep patterns, SpO2, heart rate, and glucose fluctuations.
Food Logging & Auto Scan: Uses AI-powered image recognition to scan and log food, ensuring accurate calorie and nutrient tracking.
Metabolic Analysis: Computes a metabolic score based on food intake, physical activity, and biometric data to understand how the body reacts to different foods and exercises.
Chronic Disease Prediction & Management:
- Detects early signs of diseases using symptom analysis and blood test data.
- Creates personalized health recovery plans for diabetes, obesity, cardiovascular diseases, and metabolic disorders using biohacking techniques.
3D Digital Health Twin: Generates a 3D model of your body and organs, with color-coded health indicators showing real-time metabolic and organ status.
AI-Powered Coaching:
- Suggests customized diet plans, exercises, and lifestyle modifications based on metabolic patterns.
- Provides mental health support using an LLM-driven psychiatric voice and text assistant.

How We Built It

🔬 LLM & ML Models:

Fine-tuned an LLM (Large Language Model) for health scoring.
Developed a machine learning model for food identification and metabolic analysis.

⌚ Wearable Device Integration:

Connected with Fitbit and other wearables to fetch real-time biometrics.
Overcame challenges with Fitbit API migration from OAuth 1.0 to 2.0.

🏥 Disease Prediction & Metabolism Tracking:

Designed an AI model to calculate metabolic scores and predict glucose spikes.
Integrated biohacking strategies based on real-time metabolic responses.

🏾‍⚕️ 3D Digital Twin:

Developed a 3D visualization model that dynamically updates organ health status.

Challenges We Ran Into

🚧 Accuracy of Food Recognition: Training an AI model to accurately recognize diverse food items was challenging.
📊 Metabolic Score Computation: Developing a reliable algorithm to interpret metabolic activity from multiple data sources.
❤️ Heart Rate Zones & Stress Analysis: Finding the right AI model to calculate accurate stress levels and heart rate variations.
🔐 Fitbit API Authentication: Handling the migration of Fitbit OAuth 1.0 to 2.0.

Accomplishments That We're Proud Of

✅ Successfully built a fully functional metabolic tracking system integrating wearable data, AI-based food recognition, and disease prediction models.
✅ Trained and fine-tuned multiple LLMs, including a psychiatric AI assistant to provide mental health support.
✅ Developed a 3D Digital Twin that visually represents a user’s metabolic and organ health in real time—one of the most innovative features in digital health!
✅ Achieved real-time glucose spike predictions based on food intake and physical activity patterns.

What We Learned

🧠 Deep insights into human metabolism, how food interacts with the body, and ways to manage chronic diseases effectively.
🤖 Advanced AI/ML techniques for food recognition, symptom analysis, and disease prediction.
🩺 Building and integrating LLMs into real-world healthcare applications.
⌚ Wearable technology integration and real-time biometric analysis.

What's Next for D-Twina?

🚀 Building Our Own LLM: Developing a custom disease prediction and metabolic modeling LLM to improve accuracy and efficiency.
🩸 CGM Integration: Seamlessly integrating Continuous Glucose Monitors (CGMs) like Freestyle Libre for more accurate metabolic tracking.
🎮 Simulator Mode: Introducing a simulation feature where users can test different diet and exercise scenarios to predict their metabolic responses.
🌍 Expanding Biohacking Techniques: Enhancing the biohacking recommendation engine with advanced lifestyle modifications.