Contactless Fall Detection by Means of Multiple Bioradars and Transfer Learning.

Fall detection in humans is critical in the prevention of life-threatening conditions. This is especially important for elderly people who are living alone. Therefore, automatic fall detection is one of the most relevant problems in geriatrics. Bioradiolocation-based methods have already shown their efficiency in contactless fall detection. However, there is still a wide range of areas to improve the precision of fall recognition based on view-independent concepts. In particular, in this paper, we propose an approach based on a more complex multi-channel system (three or four bioradars) in combination with the wavelet transform and transfer learning. In the experiments, we have used several radar configurations for recording different movement types. Then, for the binary classification task, a pre-trained convolutional neural network AlexNet has been fine-tuned using scalograms. The proposed systems have shown a noticeable improvement in the fall recognition performance in comparison with the previously used two-bioradar system. The accuracy and Cohen's kappa of the two-bioradar system are 0.92 and 0.86 respectively, whereas the accuracy and Cohen's kappa of the four-bioradar system are 0.99 and 0.99 respectively. The three-bioradar system's performance turned out to be in between two of the aforementioned systems and its calculated accuracy and Cohen's kappa are 0.98 and 0.97 respectively. These results may be potentially used in the design of a contactless multi-bioradar fall detection system.

Fall Detection Using Multiple Bioradars and Convolutional Neural Networks.

A lack of effective non-contact methods for automatic fall detection, which may result in the development of health and life-threatening conditions, is a great problem of modern medicine, and in particular, geriatrics. The purpose of the present work was to investigate the advantages of utilizing a multi-bioradar system in the accuracy of remote fall detection. The proposed concept combined usage of wavelet transform and deep learning to detect fall episodes. The continuous wavelet transform was used to get a time-frequency representation of the bio-radar signal and use it as input data for a pre-trained convolutional neural network AlexNet adapted to solve the problem of detecting falls. Processing of the experimental results showed that the designed multi-bioradar system can be used as a simple and view-independent approach implementing a non-contact fall detection method with an accuracy and F1-score of 99%.

Challenges and Potential Solutions of Psychophysiological State Monitoring with Bioradar Technology.

Psychophysiological state monitoring provides a promising way to detect stress and accurately assess wellbeing. The purpose of the present work was to investigate the advantages of utilizing a new unobtrusive multi-transceiver system on the accuracy of remote psychophysiological state monitoring by means of a bioradar technique. The technique was tested in laboratory conditions with the participation of 35 practically healthy volunteers, who were asked to perform arithmetic and physical workload tests imitating different types of stressors. Information about any variation in vital signs, registered by a bioradar with two transceivers, was used to detect mental or physical stress. Processing of the experimental results showed that the designed two-channel bioradar can be used as a simple and relatively easy approach to implement a non-contact method for stress monitoring. However, individual specificity of physiological responses to mental and physical workloads makes the creation of a universal stress-detector classifier that is suitable for people with different levels of stress tolerance a challenging task. For non-athletes, the proposed method allows classification of calm state/mental workload and calm state/physical workload with an accuracy of 89% and 83% , respectively, without the usage of any additional a priori information on the subject.

Automated long-term contactless temperature monitoring in animals via a thermographic camera.

The paper deals with a problem of a remote prolonged temperature monitoring of biological objects. It presents an algorithm for automatic analysis and processing of video recording from a thermographic camera. Special attention is paid to the limitation of the method taking into account the possibility of its utilizing in laboratory conditions. The proposed algorithm of video analysis has three looped stages: animal location tracking, detecting specific points in the region of interest, and estimation of temperature in these points. The presented method for measuring temperature of the biological object has the following advantage: it minimizes influence on the object of the interest, and thus allows clearer understanding of animal reaction to medication or treatment.

Estimation of a priori probabilities of sleep stages: A cycle-based approach.

This paper presents a model for the estimation of a priori probabilities of sleep epoch classes based on the epoch location in a sleep cycle. These probabilities are used as additional features for sleep stage classification based on the analysis of respiratory effort. The model was validated with data of 685 subjects selected from the Sleep Heart Health Study dataset. The model improves a base algorithm by 8 percent points and demonstrates Cohen's kappa of 0.56 ± 0.12. Our results will contribute to the development of screening tools for unobtrusive sleep structure estimation.

Bioradiolocation-based sleep stage classification.

This paper presents a method for classifying wakefulness, REM, light and deep sleep based on the analysis of respiratory activity and body motions acquired by a bioradar. The method was validated using data of 32 subjects without sleep-disordered breathing, who underwent a polysomnography study in a sleep laboratory. We achieved Cohen's kappa of 0.49 in the wake-REM-light-deep sleep classification, 0.55 for the wake-REM-NREM classification and 0.57 for the sleep/wakefulness determination. The results might be useful for the development of unobtrusive sleep monitoring systems for diagnostics, prevention, and management of sleep disorders.

Sleep architecture measurement based on cardiorespiratory parameters.

This paper presents a method for the detection of wakeful state, rapid eye movement sleep (REM), light sleep (N1&N2) and deep sleep (N3&N4) based on cardiorespiratory parameters. Experiments were conducted with data of 625 subjects without sleep-disordered breathing selected from the SHHS dataset. Compared to previous studies, our method considers results of neighboring epochs classification and epoch position over record time. The method demonstrates Cohen's kappa of 0.57 ± 0.13 and the accuracy of 71.4 ± 8.6 %. The results might contribute to the development of screening tools for diagnostics, prevention, and management of sleep disorders.

Sleep stage classification based on respiratory signal.

One of the research tasks, which should be solved to develop a sleep monitor, is sleep stages classification. This paper presents an algorithm for wakefulness, rapid eye movement sleep (REM) and non-REM sleep detection based on a set of 33 features, extracted from respiratory inductive plethysmography signal, and bagging classifier. Furthermore, a few heuristics based on knowledge about normal sleep structure are suggested. We used the data from 29 subjects without sleep-related breathing disorders who underwent a PSG study at a sleep laboratory. Subjects were directed to the PSG study due to suspected sleep disorders. A leave-one-subject-out cross-validation procedure was used for testing the classification performance. The accuracy of 77.85 ± 6.63 and Cohen's kappa of 0.59 ± 0.11 were achieved for the classifier. Using heuristics we increased the accuracy to 80.38 ± 8.32 and the kappa to 0.65 ± 0.13. We conclude that heuristics may improve the automated sleep structure detection based on the analysis of indirect information such as respiration signal and are useful for the development of home sleep monitoring system.

Sleep stage classification based on bioradiolocation signals.

This paper presents an algorithm for the detection of wakeful state, rapid eye movement sleep (REM) and non-REM sleep based on the analysis of respiratory movements acquired through a bioradar. We used the data from 29 subjects without sleep-related breathing disorders who underwent a polysomnography study at a sleep laboratory. A leave-one-subject-out cross-validation procedure was used for testing the classification performance. Cohen's kappa of 0.56 ± 0.16 and accuracy of 75.13 ± 9.81 % were achieved when compared to polysomnography results. The results of our work contribute to the development of home sleep monitoring systems.