Automated classification of hand gestures using a wristband and machine learning for possible application in pill intake monitoring.

BACKGROUND: Thanks to the increased interest towards health and lifestyle, a larger adoption in wearable devices for activity tracking is present among the general population. Wearable devices such as smart wristbands integrate inertial units, including accelerometers and gyroscopes, which can be utilised to perform automatic classification of hand gestures. This technology could also find an important application in automatic medication adherence monitoring. Accordingly, this study aims at comparing the performance of several Machine-Learning (ML) and Deep-Learning (DL) approaches for the automatic identification of hand gestures, with a specific focus on the drinking gesture, commonly associated to the action of oral intake of a pill-packed medication. METHODS: A method to automatically recognize hand gestures in daily living is proposed in this work. The method relies on a commercially available wristband sensor (MetaMotionR, MbientLab Inc.) integrating tri-axial accelerometer and gyroscope. Both ML and DL algorithms were evaluated for both multi-gesture (drinking, eating, pouring water, opening a bottle, typing, answering a phone, combing hair, and cutting) and binary gesture (drinking versus other gestures) classification from wristband sensor signals. Twenty-two participants were involved in the experimental analysis, performing a 10 min acquisition in a laboratory setting. Leave-one-subject-out cross validation was performed for robust performance assessment. RESULTS: The highest performance was achieved using a convolutional neural network with long- short term memory (CNN-LSTM), with a median f1-score of 90.5 [first quartile: 84.5; third quartile: 92.5]% and 92.5 [81.5;98.0]% for multi-gesture and binary classification, respectively. CONCLUSIONS: Experimental results showed that hand gesture classification with ML/DL from wrist accelerometers and gyroscopes signals can be performed with reasonable accuracy in laboratory settings, paving the way for a new generation of medical devices for monitoring medical adherence.

Decision Support Systems in HF based on Deep Learning Technologies.

PURPOSE OF REVIEW: Application of deep learning (DL) is growing in the last years, especially in the healthcare domain. This review presents the current state of DL techniques applied to electronic health record structured data, physiological signals, and imaging modalities for the management of heart failure (HF), focusing in particular on diagnosis, prognosis, and re-hospitalization risk, to explore the level of maturity of DL in this field. RECENT FINDINGS: DL allows a better integration of different data sources to distillate more accurate outcomes in HF patients, thus resulting in better performance when compared to conventional evaluation methods. While applications in image and signal processing for HF diagnosis have reached very high performance, the application of DL to electronic health records and its multisource data for prediction could still be improved, despite the already promising results. Embracing the current big data era, DL can improve performance compared to conventional techniques and machine learning approaches. DL algorithms have potential to provide more efficient care and improve outcomes of HF patients, although further investigations are needed to overcome current limitations, including results generalizability and transparency and explicability of the evidences supporting the process.

Feasibility of Heart Rate and Respiratory Rate Estimation by Inertial Sensors Embedded in a Virtual Reality Headset.

Virtual reality (VR) headsets, with embedded micro-electromechanical systems, have the potential to assess the mechanical heart's functionality and respiratory activity in a non-intrusive way and without additional sensors by utilizing the ballistocardiographic principle. To test the feasibility of this approach for opportunistic physiological monitoring, thirty healthy volunteers were studied at rest in different body postures (sitting (SIT), standing (STAND) and supine (SUP)) while accelerometric and gyroscope data were recorded for 30 s using a VR headset (Oculus Go, Oculus, Microsoft, USA) simultaneously with a 1-lead electrocardiogram (ECG) signal for mean heart rate (HR) estimation. In addition, longer VR acquisitions (50 s) were performed under controlled breathing in the same three postures to estimate the respiratory rate (RESP). Three frequency-based methods were evaluated to extract from the power spectral density the corresponding frequency. By the obtained results, the gyroscope outperformed the accelerometer in terms of accuracy with the gold standard. As regards HR estimation, the best results were obtained in SIT, with Rs2 (95% confidence interval) = 0.91 (0.81-0.96) and bias (95% Limits of Agreement) -1.6 (5.4) bpm, followed by STAND, with Rs2= 0.81 (0.64-0.91) and -1.7 (11.6) bpm, and SUP, with Rs2 = 0.44 (0.15-0.68) and 0.2 (19.4) bpm. For RESP rate estimation, SUP showed the best feasibility (98%) to obtain a reliable value from each gyroscope axis, leading to the identification of the transversal direction as the one containing the largest breathing information. These results provided evidence of the feasibility of the proposed approach with a degree of performance and feasibility dependent on the posture of the subject, under the conditions of keeping the head still, setting the grounds for future studies in real-world applications of HR and RESP rate measurement through VR headsets.

Evaluation of Cardiac Circadian Rhythm Deconditioning Induced by 5-to-60 Days of Head-Down Bed Rest.

Head-down tilt (HDT) bed rest elicits changes in cardiac circadian rhythms, generating possible adverse health outcomes such as increased arrhythmic risk. Our aim was to study the impact of HDT duration on the circadian rhythms of heart beat (RR) and ventricular repolarization (QTend) duration intervals from 24-h Holter ECG recordings acquired in 63 subjects during six different HDT bed rest campaigns of different duration (two 5-day, two 21-day, and two 60-day). Circadian rhythms of RR and QTend intervals series were evaluated by Cosinor analysis, resulting in a value of midline (MESOR), oscillation amplitude (OA) and acrophase (φ). In addition, the QTc (with Bazett correction) was computed, and day-time, night-time, maximum and minimum RR, QTend and QTc intervals were calculated. Statistical analysis was conducted, comparing: (1) the effects at 5 (HDT5), 21 (HDT21) and 58 (HDT58) days of HDT with baseline (PRE); (2) trends in recovery period at post-HDT epochs (R) in 5-day, 21-day, and 60-day HDT separately vs. PRE; (3) differences at R + 0 due to bed rest duration; (4) changes between the last HDT acquisition and the respective R + 0 in 5-day, 21-day, and 60-day HDT. During HDT, major changes were observed at HDT5, with increased RR and QTend intervals' MESOR, mostly related to day-time lengthening and increased minima, while the QTc shortened. Afterward, a progressive trend toward baseline values was observed with HDT progression. Additionally, the φ anticipated, and the OA was reduced during HDT, decreasing system's ability to react to incoming stimuli. Consequently, the restoration of the orthostatic position elicited the shortening of RR and QTend intervals together with QTc prolongation, notwithstanding the period spent in HDT. However, the magnitude of post-HDT changes, as well as the difference between the last HDT day and R + 0, showed a trend to increase with increasing HDT duration, and 5/7 days were not sufficient for recovering after 60-day HDT. Additionally, the φ postponed and the OA significantly increased at R + 0 compared to PRE after 5-day and 60-day HDT, possibly increasing the arrhythmic risk. These results provide evidence that continuous monitoring of astronauts' circadian rhythms, and further investigations on possible measures for counteracting the observed modifications, will be key for future missions including long periods of weightlessness and gravity transitions, for preserving astronauts' health and mission success.