This technology integrates a non-contact smart camera with deep learning models to monitor patient activity 24/7. By combining data from electronic medical records and nursing notes, it predicts pressure injury risks and provides early warnings. With over 80% prediction accuracy, the system has received invention patents in Taiwan, the US, and China. It is currently undergoing clinical validation and commercialization in collaboration with healthcare institutions.

This technology pioneers a contactless pressure injury prediction model based solely on patient activity video. Without relying on sensors or wound images, it uses Temporal Convolutional Networks (TCN) and difference map techniques to achieve AUROC around 0.80. The model enables real-time, directional risk forecasting while preserving privacy, overcoming the limitations of traditional scales. It demonstrates strong clinical potential and international competitiveness.

This technology offers a non-contact, automated, and highly scalable solution for hospitals and long-term care facilities to predict pressure injury risks and provide early warnings. It integrates smart camera hardware with AI analytics and will adopt a subscription-based model. Future extensions include fall prediction and remote care, supporting the development of exportable smart healthcare systems.