VytalLink: An AI-Powered Healthcare System That Saves Lives

Overview

Early detection of medical emergencies remains a major challenge, particularly in elderly populations, where falls account for over 37 million severe injuries annually worldwide (World Health Organization). Existing monitoring systems typically rely on either wearable sensors or vision-based tracking, which limits accuracy due to incomplete contextual understanding. Studies have shown that multimodal systems combining physiological and visual data outperform single-modality approaches in healthcare monitoring tasks (National Institutes of Health). This project introduces VytalLink, a multimodal, AI-driven health monitoring system that integrates real-time physiological signals and computer vision through a novel context fusion architecture. The system uses a Jetson Orin Nano for edge inference, processing simulated vital signs such as heart rate alongside video input for posture and activity recognition. Convolutional neural networks are used for visual detection, while time-series models analyze physiological trends for anomalies. A key innovation is the context fusion layer, which combines outputs from both modalities into a unified representation. By leveraging this approach, VytalLink detects complex events, such as falls accompanied by abnormal vitals, with greater reliability and accuracy than existing solutions. The system operates in real time on edge hardware, reducing latency and preserving privacy compared to cloud-based systems. Data is transmitted via MQTT for remote monitoring and visualization. Performance was evaluated using accuracy, precision, recall, F1 score, and latency, showing improved detection capability over baseline single-input models. Overall, VytalLink demonstrates that integrating computer vision and physiological data significantly enhances early warning systems in healthcare. This approach provides a scalable, efficient solution for proactive monitoring in both home and clinical environments, with the potential to reduce delayed intervention and improve patient outcomes.