Development of artificial intelligence edge computing based wearable device for fall detection and prevention of elderly people

Abstract

Elderly falls are a major concerning threat resulting in over 1.5-2 million elderly people experiencing severe injuries and 1 million deaths yearly. Falls experienced by Elderly people may lead to a long-term negative impact on their physical and psychological health conditions. Major healthcare research had focused on this lately to detect and prevent the fall. In this work, an Artificial Intelligence (AI) edge computing based wearable device is designed and developed for detection and prevention of fall of elderly people. Further, the various deep learning algorithms such as Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU) are utilized for activity recognition of elderly. Also, the CNN-LSTM, RNN-LSTM and GRU-LSTM with and without attention layer respectively are utilized and the performance metrics are analyzed to find the best deep learning model. Furthermore, the three different hardware boards such as Jetson Nano developer board, Raspberry PI 3 and 4 are utilized as an AI edge computing device and the best deep learning model is implemented and the computation time is evaluated. Results demonstrate that the CNN-LSTM with attention layer exhibits the accuracy, recall, precision and F1_Score of 97%, 98%, 98% and 0.98 respectively which is better when compared to other deep learning models. Also, the computation time of NVIDIA Jetson Nano is less when compared to other edge computing devices. This work appears to be of high societal relevance since the proposed wearable device can be used to monitor the activity of elderly and prevents the elderly falls which improve the quality of life of elderly people.

Keywords: Accelerometer sensor; Deep learning models; Fall detection; Internet of things; Secure pairing; Wearables.

Fig. 1

Overall Block diagram of proposed wearable device.

Fig. 2

Block diagram of proposed fall detection approach.

Fig. 3

Architecture of proposed CNN model.

Fig. 4

Basic block diagram of the proposed RNN.

Fig. 5

Basic block diagram for LSTM.

Fig. 6

Basic block diagram of GRU.

Fig. 7

Flowchart for proposed fall detection approach.

Fig. 8

Unfiltered Accelerometer data collected from elderly during various events (a) fall, (b) fall from bed, (c) lying in bed, (d) walking in uneven surface and (e) walking in even surface.

Fig. 9

Filtered Accelerometer data collected from elderly during various events (a) fall, (b) fall from bed, (c) lying in bed, (d) walking in uneven surface and (e) walking in even surface.

Fig. 10

Confusion matrix of various deep learning models (a) RNN (b) CNN (c) GRU (d) LSTM (e) CNN-LSTM (f) RNN-LSTM (g) GRU-LSTM (h) CNN-LSTM with attention (i) RNN-LSTM with attention (j) GRU-LSTM with attention.

Fig. 10

Confusion matrix of various deep learning models (a) RNN (b) CNN (c) GRU (d) LSTM (e) CNN-LSTM (f) RNN-LSTM (g) GRU-LSTM (h) CNN-LSTM with attention (i) RNN-LSTM with attention (j) GRU-LSTM with attention.

Fig. 11

Accuracy Metrics of Simple Models (CNN vs. RNN vs. GRU vs. LSTM).

Fig. 12

Accuracy Metrics of Coupled Models (CNN + LSTM vs. RNN + LSTM vs. GRU + LSTM).

Fig. 13

Accuracy Metrics of Coupled Models with Attention Layer (CNN + LSTM + ATT vs. RNN + LSTM + ATT vs. GRU + LSTM + ATT).

Fig. 14

Comparison of F1_Scores of Simple Models (CNN vs. RNN vs. GRU vs. LSTM).

Fig. 15

Comparison of F1_Scores of Coupled Models (CNN + LSTM vs. RNN + LSTM vs. GRU + LSTM).

Fig. 16

Comparison of F1_Scores of Coupled Models with Attention Layer (CNN + LSTM + ATT vs. RNN + LSTM + ATT vs. GRU + LSTM + ATT).

Fig. 17

Comparison of computation time of different AI edge computing devices.

Fig. 18

Proposed wearable device fixed to the elderly.

Fig. 19

Gui for caretaker.