Feature-based Method for Nurse Care Complex Activity Recognition from Accelerometer Sensor

As the number of patients in hospitals is increasing day by day, proper monitoring and hospital care towards patients must be ensured. In this regard, Nurse care activity recognition can play a significant role in improving the existing healthcare system. Research in this domain is very challenging because nursing activities are very complex and troublesome than other normal activities. Nursing activities are dependent not only on nurses but also on the patients' various states of illness. As a result, each activity has a high intra-class variation. We have participated in '3rd Nurse Care Activity Recognition Challenge 2021' and proposed a simple machine learning approach to recognize nursing activities. After data pre-processing and feature engineering, we have used several machine learning algorithms. Among them, we have achieved our best results in the Random Forest model. Using this model, We have obtained 72 percent validation accuracy classifying several challenging activities.