Spatio-Temporal Transformer with Hypergraph in Nursing Activity Recognition.

Area of Science:

• Computer Science
• Artificial Intelligence
• Healthcare Technology

Background:

• The growing elderly population necessitates increased demand for nursing care.
• Manual record-keeping by nurses is time-consuming and prone to errors.
• Accurate recognition of nursing activities can support nurse training and robotic development.

Purpose of the Study:

• To develop an automated system for recognizing nursing actions.
• To leverage skeletal data for accurate activity recognition.
• To reduce the burden of manual data recording for nurses.

Main Methods:

• A novel hypergraph transformer-based network was developed.
• The network extracts features from skeletal joint data.
• The approach focuses solely on utilizing skeletal data for recognition.

Main Results:

• The proposed method achieved a state-of-the-art accuracy of 79.3%.
• The system demonstrated effectiveness using only skeletal data.
• The results were validated on the Nurse Care Activity Recognition Challenge dataset.

Conclusions:

• Automated nursing action recognition using skeletal data is feasible and accurate.
• This technology can enhance nurse training and facilitate the development of nursing robots.
• The hypergraph transformer network shows promise for future healthcare applications.