i-Care
AI-Powered Patient Monitoring System
Leveraging YOLO-based pose estimation for continuous, real-time surveillance in hospitals and ICUs. Detects falls, abnormal movements, and unsafe postures — delivering instant alerts and a centralized clinical dashboard to reduce caregiver workload and enhance patient safety.
Vision AI That Watches Over Your Patients
i-Care transforms any standard CCTV or IP camera into an intelligent patient guardian. By analyzing skeletal keypoints in real time, the system understands exactly what a patient is doing — and immediately raises the alarm when something is wrong.
Unlike wearable-based solutions, i-Care requires zero patient compliance. There's nothing to wear, charge, or forget. One camera. Continuous protection.
- Edge inference — no data leaves the premises
- Works on existing hospital CCTV infrastructure
- Centralized dashboard for multi-room monitoring
- Role-based access: doctor, nurse, admin
What i-Care Detects
Six layers of AI-powered detection working simultaneously on a single camera stream.
Real-Time Pose Estimation
YOLO-Pose skeletal keypoint analysis detects head, torso, and limb positions frame-by-frame at 20–30 FPS per camera stream.
Fall Detection
Instant alerts when a patient falls — seconds of response time saved can be life-critical in ICU or post-operative recovery.
Abnormal Movement Detection
Identifies unsafe or unexpected movements including reaching, rolling off-bed, or sudden convulsions with immediate alarm.
Activity Recognition
Classifies patient states — lying, sitting, standing, walking — enabling automated care logs without manual entry.
Zone-Based Logic
Tracks patient entry and exit in defined zones (washroom, bed area) for context-aware event classification.
Edge + Cloud Hybrid
Runs inference on edge devices (Jetson Orin / RPi 5) with optional server aggregation for multi-room, multi-patient deployments.
Technical Architecture
Hardware and software stack powering i-Care's real-time AI pipeline.
- Edge Device (Recommended)
- NVIDIA Jetson Orin Nano / Orin NX
- Alternative Edge
- RPi 5 + Hailo / Coral AI Accelerator
- Inference Speed
- 20–30 FPS per stream
- Camera
- IP / USB · min 1080p @ 30 FPS
- Lens
- Wide-angle (ICU full-coverage)
- Server (Multi-Room)
- Intel i7 / Ryzen 7 · RTX 3060+
- RAM
- 16–32 GB (server setup)
- Storage
- SSD (event image & log archive)
- AI Model
- YOLO Pose (custom-tuned)
- Backend
- Python · OpenCV · PyTorch
- API Layer
- FastAPI / Flask
- Dashboard
- Web-based · Live video preview
- Alerts
- Sound + push notification
- Access Control
- Role-based (doctor / nurse / admin)
- Multi-Camera
- Supported
- Data
- Event snapshots + timeline logs
Real-World Applications
Hospitals & ICUs
Continuous monitoring of sedated, post-operative, or high-risk patients. Instant fall detection and self-harm prevention without increasing nurse workload.
Elderly Care Centers
24/7 non-intrusive visual monitoring for senior residents — reducing dependency on constant physical supervision while maintaining dignity.
Home Healthcare
Remote patient monitoring with caregiver notification. Ideal for bedridden patients recovering at home with minimal on-site presence.
Medical Research
Longitudinal patient behavior analysis, recovery pattern monitoring, and anonymised clinical data generation for research insights.
See i-Care In Action
Watch the live pose estimation and event detection demo.
i-Care Version 02
The next evolution fuses visual AI with physiological sensing — merging YOLO pose data with real-time biometrics from a smart patient bracelet for near-perfect incident prediction.
IoT Patient Bracelet
Heart rate, SpO₂, and accelerometer data fused with visual pose estimation for unprecedented monitoring accuracy.
Predictive AI Alerts
ML models trained on historical event data to anticipate falls and critical incidents seconds before they happen.
Clinical Analytics Suite
Aggregate patient behaviour data into clinical dashboards for doctor review, audit trails, and outcome research.
i-Care in Action
Real detection outputs from live hospital environments.
Deploy i-Care in Your Facility
Works with your existing cameras. No wearables required. Get a custom deployment proposal tailored to your ward layout and patient monitoring requirements.