Health Monitoring for Stroke Patients Using Machine Learning

Abstract - Stroke patients often face challenges in communication and independent medication management, necessitating innovative assistive technologies for better healthcare monitoring. This project introduces a real-time hand gesture-based system to monitor tablet intake in stroke patients using computer vision and machine learning techniques. The solution integrates OpenCV for video processing, Media Pipe for accurate hand landmark detection, and Twilio for sending automated SMS alerts to caregivers. The system operates by detecting specific hand gestures-particularly the movement of the index finger toward the mouth-which signifies an attempt to consume medication. When this gesture is confirmed, the system logs the event, updates the tablet count, and triggers both a visual alert using Tkinter and an SMS notification via Twilio. This ensures real-time tracking and caregiver awareness without physical intervention. This low-cost, non-invasive solution can be implemented using a standard webcam and basic hardware, making it especially suitable for home healthcare environments. The methodology enhances patient autonomy while maintaining safety and accountability. Future developments may incorporate deep learning for more robust gesture recognition and extend the system’s functionality to cover broader health monitoring scenarios. Key Words: Hand Gesture Recognition, Stroke Patient Monitoring, Computer Vision, Real-time Alert System. .