The Institute for Information Industry (III) has unveiled a new 5G and AI-enabled smart mattress system that can predict when elderly individuals will get out of bed up to 90 seconds in advance. The new technology aims to reduce falls in long-term care facilities and ease caregiver stress. The system has earned global recognition and awards for corporate social responsibility.
Li Ying-fang (李穎芳) from III says that elderly people, particularly those with dementia, often fear getting up from bed without realizing it or may attempt to get up by themselves to avoid bothering others. Therefore, predicting their behavior in advance is crucial. Li says the system integrates Compal Electronics’ 5G tech with Humetrics’ smart mattress to provide early warnings to caregivers. This advance notice helps prevent falls and minimizes the need for frequent room checks.
Li says the model is tailored for dementia patients, detecting pre-get-up patterns that might otherwise go unnoticed. It takes about two days for the mattress to learn a patient’s restlessness model fully. While the general model has a 70% accuracy rate, personal predictions improve significantly with an additional two days of learning.
The system is currently implemented at the Taichung Yongxin Xinjia Community long-term care facility and has benefited over 5,400 people. Rental costs for the mattress are around NT$1,500 (US$47) per month and can be used with home wireless networks, aiming to make it widely accessible and alleviate caregiver stress.