Human Activity Recognition Using Deep Learning Techniques for Healthcare Applications
Raj Shekhar, Deepak Singh Tomar, Bhagwati Sharan, Rajesh Kumar Pateriya · 2024
The idea of smart healthcare is gradually gaining traction with the rapid advancement in information technologies. Smart healthcare is the intelligent transformation of the current medical system to make it more reliable, efficient, and individualized through the intelligent use of next-generation technologies like artificial intelligence, and the Internet of Things (IoT). This study offered the recent advancement of Deep learning (DL) techniques for healthcare systems, and the use of DL techniques to identify human physical activity with wearable sensors. In this paper, convolutional neural networks (CNN), CNN-LSTM, Multi Headed CNN-LSTM, and Multi Headed CNNBiLSTM are used. The obtained outcomes of these Deep Learning models are compared in terms of accuracy, precision, recall and F1 score. The Multi-Headed CNN-BiLSTM model exhibits better performance than the rest of the models when applied to the UCI HAR dataset.