Recent advances in artificial intelligence and machine learning (ML) led to effective methods and tools for analyzing the human behavior. Human Activity Recognition (HAR) is one of the fields that has seen an explosive research interest among the ML community due to its wide range of applications. HAR is one of the most helpful technology tools to support the elderly’s daily life and to help people suffering from cognitive disorders, Parkinson’s disease, dementia, etc. It is also very useful in areas such as transportation, robotics and sports. Deep learning (DL) is a branch of ML based on complex Artificial Neural Networks (ANNs) that has demonstrated a high level of accuracy and performance in HAR. Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) are two types of DL models widely used in the recent years to address the HAR problem. The purpose of this paper is to investigate the effectiveness of their integration in recognizing daily activities, e.g., walking. We analyze four hybrid models that integrate CNNs with four powerful RNNs, i.e., LSTMs, BiLSTMs, GRUs and BiGRUs. The outcomes of our experiments on the PAMAP2 dataset indicate that our proposed hybrid models achieve an outstanding level of performance with respect to several indicative measures, e.g., F-score, accuracy, sensitivity, and specificity.

Background: The tracking devices could help measuring the heart rate and energy expenditure and recognizing the user's activity. The calorie measurement is a significant achievement for the fitness tracking and the continuous health monitoring. Methods: In this paper, a combination of an accelerometer and a photoplethysmography (PPG) sensor is implemented to calculate the calories consumed. These sensors were mounted next to each other and then were placed on the ankle and finger by flat cable. The sensed data are transferred via Bluetooth to a smartphone in a serial and real-time manner. An Android App is designed to display the user's health data. The average amount of consumed energy is obtained from the combination of the accelerometer sensor based on the laws of motion and the PPG sensor based on the heart rate data. Results: The designed system is tested on 10 nonathlete males and 10 nonathlete females randomly. By applying thevelet, the value of the acceleration signal variance was reduced from 3.2 to 0.8. The correlation between PPG and pulse oximeter was 0.9. Moreover, the correlation of the accelerometer and treadmill was 0.9. The root mean square error (RMSE) and the P value of the calorie output from PPG and pulse oximeter are 0.53 and 0.008, respectively. The RMSE and the P value of the calories output from the accelerometer and the treadmill are 0.42 and 0.007, respectively. Conclusion: Our device validity and reliability were good by comparing it with a typical smart band, smart watch, and smartphone available in the market. The combined PPG and the accelerometer sensors were compared with the gold standard, the pulse oximeter, and the treadmill. According to the results, there is no significant difference in the values obtained. Therefore, a mobile system is augmented with the wireless accelerometer and PPG that are connected to a smartphone. The system could be carried out with the user at any time and any place.