Validation of human activity recognition using a convolutional neural network on accelerometer and gyroscope data

Single Accelerometer to Recognize Human Activities Using Neural Networks

Exoskeletons have decreased physical effort and increased comfort in activities of daily living (ADL) such as walking, squatting, and running. However, this assistance is often activity specific and does not accommodate a wide variety of different activities. To overcome this limitation and increase the scope of exoskeleton application, an automatic human activity recognition (HAR) system is necessary. We developed two deep-learning models for HAR using one-dimensional-convolutional neural network (CNN) and a hybrid model using CNNs and long-short term memory (LSTM). We trained both models using the data collected from a single three-axis accelerometer placed on the chest of ten subjects. We were able to classify five different activities, standing, walking on level ground, walking on an incline, running, and squatting, with an accuracy of 98.1% and 97.8%, respectively. A two subject real-time validation trial was also conducted to validate the real-time applicability of the system. The real-time accuracy was measured at 96.6% and 97.2% for the CNN and the hybrid model, respectively. The high classification accuracy in the test and real-time evaluation suggests that a single sensor could distinguish human activities using machine-learning-based models.

Efficacy of Individualized Sensory-Based mHealth Interventions to Improve Distress Coping in Healthcare Professionals: A Multi-Arm Parallel-Group Randomized Controlled Trial

Detrimental effects of chronic stress on healthcare professionals have been well-established, but the implementation and evaluation of effective interventions aimed at improving distress coping remains inadequate. Individualized mHealth interventions incorporating sensor feedback have been proposed as a promising approach. This study aimed to investigate the impact of individualized, sensor-based mHealth interventions focusing on stress and physical activity on distress coping in healthcare professionals. The study utilized a multi-arm, parallel group randomized controlled trial design, comparing five intervention groups (three variations of web-based training and two variations of an app training) that represented varying levels of individualization to a control group. Both self-reported questionnaire data (collected using Limesurvey) as well as electrocardiography and accelerometry-based sensory data (collected using Mesana Sensor) were assessed at baseline and post-intervention (after eight weeks). Of the 995 eligible participants, 170 (26%) completed the post-intervention measurement (Group 1: N = 21; Group 2: N = 23; Group 3: N = 7; Group 4: N = 34; Group 5: N = 16; Control Group: N = 69). MANOVA results indicated small to moderate time-by-group interaction effects for physical activity-related outcomes, including moderate to vigorous physical activity (F(1,5) = 5.8, p = ≤0.001, η2p = 0.057) and inactivity disruption (F(1,5) = 11.2, p = <0.001, η2p = 0.100), in the app-based intervention groups, but not for step counts and inactivity. No changes were observed in stress-related heart rate variability parameters over time. Despite a high dropout rate and a complex study design, the individualized interventions showed initial positive effects on physical activity. However, no significant changes in stress-related outcomes were observed, suggesting that the intervention duration was insufficient to induce physiological adaptations that would result in improved distress coping.

An Intelligent Cost-Efficient System to Prevent the Improper Posture Hazards in Offices Using Machine Learning Algorithms

In this research, an intelligent and cost-efficient system has been proposed to detect the improper sitting posture of a person working at a desk, mostly in offices, using machine learning classification techniques. The current era demands to avoid the harms of an improper posture as it, when prolonged, is very painful and can be fatal sometimes. This study also includes a comparison of two arrangements. Arrangement 01 includes six force-sensitive resistor (FSR) sensors alone, and it is less expensive. Arrangement 02 consists of two FSR sensors and one ultrasonic sensor embedded in the back seat of a chair. The K-nearest neighbor (KNN), Naive Bayes, logistic regression, and random forest algorithms are used to augment the gain and enhanced accuracy for posture detection. The improper postures recognized in this study are backward-leaning, forward-leaning, left-leaning, and right-leaning. The presented results validate the proposed system as the accuracy of 99.8% is achieved using a smaller number of sensors that make the proposed prototype cost-efficient with improved accuracy and lower execution time. The proposed model is of a dire need for employees working in offices or even at the residential level to make it convenient to work for hours without having severe effects of improper posture and prolonged sitting.

Measuring catatonia motor behavior with objective instrumentation

OBJECTIVE

Catatonia is a neuropsychiatric syndrome, with important psychomotor features, associated with schizophrenia and other psychiatric disorders. The syndrome comprises multiple symptoms including abnormal motor control, behaviors, volition, and autonomic regulation. Catatonia assessment relies on clinical rating scales and clinicians familiar with the catatonia exam. However, objective instrumentation may aid the detection of catatonia. We aimed to investigate the relationship between movement parameters derived from actigraphy and expert ratings of catatonia symptoms measured by the Bush Francis Catatonia Rating Scale (BFCRS) and the Northoff Catatonia scale (NCS).

METHODS

Eighty-six acutely ill inpatients with schizophrenia spectrum disorders were assessed with the BFCRS, the NCS, and 24 h continuous actigraphy. Non-wear and sleep periods were removed from the actigraphy data prior to analysis. Associations between total catatonia scores, derived from both BFCRS and NCS, and actigraphy parameters as well as between single BFCRS items and actigraphy parameters were calculated using Spearman's rank correlation and non-parametric ANCOVAs (Quade's ANCOVAs), respectively.

RESULTS

Both higher BFCRS total scores (r = 0.369, p = 0.006) and NCS total scores (r = 0.384, p = 0.004) were associated with lower activity levels (AL). Higher scores on single BFCRS items such as immobility/stupor or staring were linked to lower AL (immobility/stupor: F = 17.388, p < 0.001, η² = 0.175; staring: F = 7.849, p = 0.001, η² = 0.162) and lower metabolic equivalents of task (MET).

CONCLUSION

Specific catatonia symptoms such as immobility/stupor and staring can be measured with actigraphy. This may aid the detection, staging, and monitoring of catatonia in clinical settings.