Flutter Mobile Layer
The companion app is built with Flutter to deliver cross-platform role-based flows for Patient, Caregiver, and Doctor users.
System Architecture
The system combines a Flutter mobile app, Flask backend services, and AWS cloud infrastructure to keep the wearable experience lightweight while the heavy work happens remotely.
The companion app is built with Flutter to deliver cross-platform role-based flows for Patient, Caregiver, and Doctor users.
Core APIs are implemented with Flask for authentication, profile management, AI endpoint routing, and secure token-based access.
AWS powers compute, storage, and database services using EC2, S3, and RDS to handle heavy processing reliably at scale.
Raspberry Pi + camera capture real-time visual context and send only the required data stream for processing.
Face recognition and medicine/object detection execute in controlled modes to maintain low latency and stable response.
Role-based chatbot and clinical support logic enrich raw detections into safe, understandable guidance.
Flask APIs with JWT validation, rate limiting, and standardized error handling enforce reliable operations.
AWS EC2, S3, and RDS ensure model availability, scalable processing, and persistent medical data integrity.
Retraining pipelines ingest new faces from mobile uploads and hot-update active models without downtime.
Real-time image processing from AR glasses: face recognition via SVM, object detection via YOLOv8, with immediate OLED output.
Multimodal RAG system with Gemini: patient questions, STT/TTS, database context, and ResNet50 MRI analysis for doctors.
Caregiver uploads images to S3, the training server refreshes the SVM model, and the main API hot reloads the latest version.
JWT-based stateless auth with token blacklist, password signature validation, and role-based access control.