Synulin AI: Synchronized Insulin Pen with Integrated Al Agent Control (2024-2027). Reference: CPP2024-011898. Funded by AEI, total: 875.000,00 € P.I. Josep Vehí Casellas. Other participants: Insulcloud, CLÍNIC BARCELONA Hospital general de segovia.
SUMMARY:
Synulin Al integrates a smart insulin pen, a VHT, an Al agent, and gamification to address critical challenges in insulin-dependent diabetes via a multidisciplinary approach (biomedical engineering, data science, health management). The smart pen/Insulclock platform provides accurate dose logging, sensors, connectivity, and a compact, eco-efficient design forming the data backbone for intelligence and coaching.
Bridges VHT and the serious game to deliver personalized training and recommendations, acting proactively (when detecting risk/inactivity).
Continuous glycemic monitoring; alerts/reminders to sustain engagement; semi-automatic therapy adjustments from VHT simulations; behavior tracking (meals, exercise, routines); lifestyle coaching with negotiated goals and game-based incentives.
Large Language Models (LLM) and Retrieval Augmented Generation (RAG)
stack: Retrieves patient-specific context and fuses it with HVT simulations and real-time data to produce precise, compliant guidance, preventing cross-patient data leakage.
Main Goal
MAIN GOAL:
The main objective of Synulin Al is to transform insulin-dependent diabetes management through a synchronized, Al agent-controlled insulin pen, supported by a Virtual Human Twin (VHT) and gamification, with clinical validation. This goal will be achieved through the following objectives:
Automated, Al-agent-controlled pen; miniaturization, lifetime power, sensors and connectivity for dose logging and safety alerts.
Patient segmentation and predictive maintenance to personalize therapy, detect patterns/failures, and increase robustness.
Scenario simulations (meals, exercise, dosing) and gamification to enhance education, engagement, and adherence.
Design of an Al agent for therapeutic guidance and coaching.
Serious game and adherence: Translates guidance into missions, points, and rewards to reinforce understanding, engagement, and long-term adherence.
TEAM:
MICELAB:
Research leads:
- Josep Vehí Casellas
- Iván Contreras
- Imma Boada
EXPECTED RESULTS:
Development of two innovative devices to improve insulin delivery, an Al-enabled smart insulin pen and the Insulclock v2.1 cap.
VHT and Al coaching agent: Patient-specific simulations fused with real-time data to deliver personalized therapy guidance and proactive alerts (dosing, hypoglycemia), strengthening decision-making, adherence, and overall outcomes.
Serious-game integration creates an interactive, risk-free learning environment that deepens diabetes understanding, promotes healthier habits, and supports adherence.
Greater public access to advanced diabetes-management hardware is expected, along with higher patient engagement and participation in their care, improving diabetes management through advanced data analytics and predictive maintenance.