Search

Wearable activity sensors typically count movement quantity, such as the number of steps taken or the number of upper extremity (UE) counts achieved. However, for some applications, such as neurologic rehabilitation, it may be of interest to quantify the quality of the movement experience (QOME), defined, for example, as how diverse or how complex movement epochs are. We previously found that individuals with UE impairment after stroke exhibited differences in their distributions of forearm postures across the day and that these differences could be quantified with kurtosis-an established statistical measure of the peakedness of distributions. In this paper, we describe further progress toward the goal of providing real-time feedback to try to help people learn to modulate their movement diversity. We first asked the following: to what extent do different movement activities induce different values of kurtosis? We recruited seven unimpaired individuals and evaluated a set of 12 therapeutic activities for their forearm postural diversity using kurtosis. We found that the different activities produced a wide range of kurtosis values, with conventional rehabilitation therapy exercises creating the most spread-out distribution and cup stacking the most peaked. Thus, asking people to attempt different activities can vary movement diversity, as measured with kurtosis. Next, since kurtosis is a computationally expensive calculation, we derived a novel recursive algorithm that enables the real-time calculation of kurtosis. We show that the algorithm reduces computation time by a factor of 200 compared to an optimized kurtosis calculation available in SciPy, across window sizes. Finally, we embedded the kurtosis algorithm on a commercial smartwatch and validated its accuracy using a robotic simulator that wore the smartwatch, emulating movement activities with known kurtosis. This work verifies that different movement tasks produce different values of kurtosis and provides a validated algorithm for the real-time calculation of kurtosis on a smartwatch. These are needed steps toward testing QOME-focused, wearable rehabilitation.

Estimation of temporospatial clinical features of gait (CFs), such as step count and length, step duration, step frequency, gait speed, and distance traveled, is an important component of community-based mobility evaluation using wearable accelerometers. However, accurate unsupervised computerized measurement of CFs of individuals with Duchenne muscular dystrophy (DMD) who have progressive loss of ambulatory mobility is difficult due to differences in patterns and magnitudes of acceleration across their range of attainable gait velocities. This paper proposes a novel calibration method. It aims to detect steps, estimate stride lengths, and determine travel distance. The approach involves a combination of clinical observation, machine-learning-based step detection, and regression-based stride length prediction. The method demonstrates high accuracy in children with DMD and typically developing controls (TDs) regardless of the participants level of ability. Fifteen children with DMD and fifteen TDs underwent supervised clinical testing across a range of gait speeds using 10 m or 25 m run/walk (10 MRW, 25 MRW), 100 m run/walk (100 MRW), 6-min walk (6 MWT), and free-walk (FW) evaluations while wearing a mobile-phone-based accelerometer at the waist near the bodys center of mass. Following calibration by a trained clinical evaluator, CFs were extracted from the accelerometer data using a multi-step machine-learning-based process and the results were compared to ground-truth observation data. Model predictions vs. observed values for step counts, distance traveled, and step length showed a strong correlation (Pearsons r = -0.9929 to 0.9986, p < 0.0001). The estimates demonstrated a mean (SD) percentage error of 1.49% (7.04%) for step counts, 1.18% (9.91%) for distance traveled, and 0.37% (7.52%) for step length compared to ground-truth observations for the combined 6 MWT, 100 MRW, and FW tasks. Our study findings indicate that a single waist-worn accelerometer calibrated to an individuals stride characteristics using our methods accurately measures CFs and estimates travel distances across a common range of gait speeds in both DMD and TD peers.

Overnight sleep staging is an important part of the diagnosis of various sleep disorders. Polysomnography is the gold standard for sleep staging, but less-obtrusive sensing modalities are of emerging interest. Here, we developed and validated an algorithm to perform “proxy” sleep staging using cardiac and respiratory signals derived from a chest-worn accelerometer. We collected data in two sleep centers, using a chest-worn accelerometer in combination with full PSG. A total of 323 participants were analyzed, aged 13–83 years, with BMI 18–47 kg/m2. We derived cardiac and respiratory features from the accelerometer and then applied a previously developed method for automatic cardio-respiratory sleep staging. We compared the estimated sleep stages against those derived from PSG and determined performance. Epoch-by-epoch agreement with four-class scoring (Wake, REM, N1+N2, N3) reached a Cohen’s kappa coefficient of agreement of 0.68 and an accuracy of 80.8%. For Wake vs. Sleep classification, an accuracy of 93.3% was obtained, with a sensitivity of 78.7% and a specificity of 96.6%. We showed that cardiorespiratory signals obtained from a chest-worn accelerometer can be used to estimate sleep stages among a population that is diverse in age, BMI, and prevalence of sleep disorders. This opens up the path towards various clinical applications in sleep medicine.

There is limited validation for portable methods in evaluating high-speed running biomechanics, with inertial measurement unit (IMU) systems commonly used as wearables for this purpose. This study aimed to evaluate the validity of an IMU system in high-speed running compared to a 3D motion analysis system (MAS). One runner performed incremental treadmill running, from 12 to 18 km/h, on two separate days. Sagittal angles for the shank, knee, hip and pelvis were measured simultaneously with three IMUs and the MAS at the point of contact (POC), the timing when the foot initially hits the ground, as identified by IMU system acceleration, and compared to the POC identified via force plate. Agreement between the systems was evaluated using intra-class correlation coefficients, Pearson’s r, Bland–Altman limits of agreements, root mean square error and paired t-tests. The IMU system reliably determined POC (which subsequently was used to calculate stride time) and measured hip flexion angle and anterior pelvic tilt accurately and consistently at POC. However, it displayed inaccuracy and inconsistency in measuring knee flexion and shank angles at POC. This information provides confidence that a portable IMU system can aid in establishing baseline running biomechanics for performance optimisation, and/or inform injury prevention programs.

Rumination behavior in cattle can provide valuable information for monitoring health status and animal welfare, but continuous monitoring is essential to detect changes in rumination behavior. In a previous study validating the use of a respiration rate sensor equipped with a triaxial accelerometer, the regurgitation process was also clearly visible in the pressure and accelerometer data. The aim of the present study, therefore, was to measure the individual lengths of rumination cycles and to validate whether the sensor data showed the same number of regurgitations as those counted visually (video or direct observation). For this purpose, 19 Holstein Friesian cows equipped with a respiration rate sensor were observed for two years, with a focus on rumination behavior. The results showed a mean duration of 59.27 ± 9.01 s (mean ± SD) per rumination cycle and good agreement (sensitivity: 99.1–100%, specificity: 87.8–95%) between the two methods (sensor and visual observations). However, the frequency of data streaming (continuously or every 30 s) from the sensor to the data storage system strongly influenced the classification performance. In the future, an algorithm and a data cache will be integrated into the sensor to provide rumination time as an additional output.

Walking is crucial for independence and quality of life. This study leverages wrist-worn sensor data from UK Biobank participants to establish normative daily-life walking data, stratified by age and sex, to provide benchmarks for research and clinical practice. The Watch Walk digital biomarkers were developed, validated, and applied to 92,022 participants aged 45–79 who wore a wrist sensor for at least three days. Normative data were collected for daily-life walking speed, step-time variability, step count, and 17 other gait and sleep biomarkers. Test–retest reliability was calculated, and associations with sex, age, self-reported walking pace, and mobility problems were examined. Population mean maximal and usual walking speeds were 1.49 and 1.15 m/s, respectively. The daily step count was 7749 steps, and step regularity was 65%. Women walked more regularly but slower than men. Walking speed, step count, longest walk duration, and step regularity decreased with age. Walking speed is associated with sex, age, self-reported pace, and mobility problems. Test–retest reliability was good to excellent (ICC ≥ 0.80). This study provides large-scale normative data and benchmarks for wrist-sensor-derived digital gait and sleep biomarkers from real-world data for future research and clinical applications.

Background: Seniors wearing a passive hip exoskeleton (Exo) show increased walking speed and step length but reduced cadence. We assessed the test–retest reliability of seniors’ gait characteristics with Exo. Methods: Twenty seniors walked with and without Exo (noExo) on a 10 m indoor track over two sessions separated by one week. Speed, step length, cadence and step time variability were extracted from one inertial measurement unit (IMU) placed over the L5 vertebra. Relative and absolute reliability were assessed using the intraclass correlation coefficient (ICC), standard error of measurement (SEM) and minimal detectable change (MDC). Results: The relative reliability of speed, step length, cadence and step time variability ranged from “almost perfect to substantial” for Exo and noExo with ICC values between 0.75 and 0.87 and 0.60 and 0.92, respectively. The SEM and MDC values for speed, step length cadence and step time variability during Exo and noExo were

Wearable activity sensors typically count movement quantity, such as the number of steps taken or the number of upper extremity (UE) counts achieved. However, for some applications, such as neurologic rehabilitation, it may be of interest to quantify the quality of the movement experience (QOME), defined, for example, as how diverse or how complex movement epochs are. We previously found that individuals with UE impairment after stroke exhibited differences in their distributions of forearm postures across the day and that these differences could be quantified with kurtosis—an established statistical measure of the peakedness of distributions. In this paper, we describe further progress toward the goal of providing real-time feedback to try to help people learn to modulate their movement diversity. We first asked the following: to what extent do different movement activities induce different values of kurtosis? We recruited seven unimpaired individuals and evaluated a set of 12 therapeutic activities for their forearm postural diversity using kurtosis. We found that the different activities produced a wide range of kurtosis values, with conventional rehabilitation therapy exercises creating the most spread-out distribution and cup stacking the most peaked. Thus, asking people to attempt different activities can vary movement diversity, as measured with kurtosis. Next, since kurtosis is a computationally expensive calculation, we derived a novel recursive algorithm that enables the real-time calculation of kurtosis. We show that the algorithm reduces computation time by a factor of 200 compared to an optimized kurtosis calculation available in SciPy, across window sizes. Finally, we embedded the kurtosis algorithm on a commercial smartwatch and validated its accuracy using a robotic simulator that “wore” the smartwatch, emulating movement activities with known kurtosis. This work verifies that different movement tasks produce different values of kurtosis and provides a validated algorithm for the real-time calculation of kurtosis on a smartwatch. These are needed steps toward testing QOME-focused, wearable rehabilitation.

The utilization of inertial measurement units as wearable sensors is proliferating across various domains, such as health care, sports, and rehabilitation. This expansion has produced a market of devices tailored to accommodate very specific ranges of operational demands. Simultaneously, this growth is creating opportunities for the development of a new class of devices more oriented towards general-purpose use and capable of capturing both high-frequency signals for short-term, event-driven motion analysis and low-frequency signals for extended monitoring. For such a design, which combines flexibility and low cost, a rigorous evaluation of the device in terms of deviation, noise levels, and precision is essential. This evaluation is crucial for identifying potential improvements and refining the design accordingly, yet it is rarely addressed in the literature. This paper presents the development process of such a device. The results of the design process demonstrate acceptable performance in optimizing energy consumption and storage capacity while highlighting the most critical optimizations needed to advance the device towards the goal of a smart, general-purpose unit for human motion monitoring.

This study aims to integrate a convolutional neural network (CNN) and the Random Forest Model into a rehabilitation assessment device to provide a comprehensive gait analysis in the evaluation of movement disorders to help physicians evaluate rehabilitation progress by distinguishing gait characteristics under different walking modes. Equipped with accelerometers and six-axis force sensors, the device monitors body symmetry and upper limb strength during rehabilitation. Data were collected from normal and abnormal walking groups. A knee joint limiter was applied to subjects to simulate different levels of movement disorders. Features were extracted from the collected data and analyzed using a CNN. The overall performance was scored with Random Forest Model weights. Significant differences in average acceleration values between the moderately abnormal (MA) and severely abnormal (SA) groups (without vehicle assistance) were observed (p < 0.05), whereas no significant differences were found between the MA with vehicle assistance (MA-V) and SA with vehicle assistance (SA-V) groups (p > 0.05). Force sensor data showed good concentration in the normal walking group and more scatter in the SA-V group. The CNN and Random Forest Model accurately recognized gait conditions, achieving average accuracies of 88.4% and 92.3%, respectively, proving that the method mentioned above provides more accurate gait evaluations for patients with movement disorders.

Accelerometers are mainly used to measure the non-conservative forces at the center of mass of gravity satellites and are the core payloads of gravity satellites. All kinds of disturbances in the satellite platform and the environment will affect the quality of the accelerometer data. This paper focuses on the quality assessment of accelerometer data from the GRACE-FO satellites. Based on the ACC1A data, we focus on the analysis of accelerometer data anomalies caused by various types of disturbances in the satellite platform and environment, including thruster spikes, peaks, twangs, and magnetic torque disturbances. The data characteristics and data accuracy of the accelerometer in different operational states and satellite observation modes are analyzed using accelerometer observation data from different time periods. Finally, the data consistency of the accelerometer is analyzed using the accelerometer transplantation method. The results show that the amplitude spectral density of three-axis linear acceleration is better than the specified accuracy (above 10−1 Hz) in the accelerometer’s nominal status. The results are helpful for understanding the characteristics and data accuracy of GRACE-FO accelerometer observations.

Landslides are a global and frequent natural hazard, affecting many communities and infrastructure networks. Technological solutions are needed for long-term, large-scale condition monitoring of infrastructure earthworks or natural slopes. However, current instruments for slope stability monitoring are often costly, require a complex installation process and/or data processing schemes, or have poor resolution. Wireless sensor networks comprising low-power, low-cost sensors have been shown to be a crucial part of landslide early warning systems. Here, we present the development of a novel sensing approach that uses linear arrays of three-axis accelerometers for monitoring changes in sensor inclination, and thus the surrounding soil's deformation. By combining these deformation measurements with depth-resolved temperature measurements, we can link our data to subsurface thermal-hydrological regimes where relevant. In this research, we present a configuration of cascaded I2C sensors that (i) have ultra-low power consumption and (ii) enable an adjustable probe length. From an electromechanical perspective, we developed a novel board-to-board connection method that enables narrow, semi-flexible sensor arrays and a streamlined assembly process. The low-cost connection method relies on a specific FR4 printed circuit board design that allows board-to-board press fitting without using electromechanical components or solder connections. The sensor assembly is placed in a thin, semi-flexible tube (inner diameter 6.35 mm) that is filled with an epoxy compound. The resulting sensor probe is connected to an AA-battery-powered data logger with wireless connectivity. We characterize the system's electromechanical properties and investigate the accuracy of deformation measurements. Our experiments, performed with probes up to 1.8 m long, demonstrate long-term connector stability, as well as probe mechanical flexibility. Furthermore, our accuracy analysis indicates that deformation measurements can be performed with a 0.390 mm resolution and a 95% confidence interval of ±0.73 mm per meter of probe length. This research shows the suitability of low-cost accelerometer arrays for distributed soil stability monitoring. In comparison with emerging low-cost measurements of surface displacement, our approach provides depth-resolved deformation, which can inform about shallow sliding surfaces.

Inertial signals are the most widely used signals in human activity recognition (HAR) applications, and extensive research has been performed on developing HAR classifiers using accelerometer and gyroscope data. This study aimed to investigate the potential enhancement of HAR models through the fusion of biological signals with inertial signals. The classification of eight common low-, medium-, and high-intensity activities was assessed using machine learning (ML) algorithms, trained on accelerometer (ACC), blood volume pulse (BVP), and electrodermal activity (EDA) data obtained from a wrist-worn sensor. Two types of ML algorithms were employed: a random forest (RF) trained on features; and a pre-trained deep learning (DL) network (ResNet-18) trained on spectrogram images. Evaluation was conducted on both individual activities and more generalized activity groups, based on similar intensity. Results indicated that RF classifiers outperformed corresponding DL classifiers at both individual and grouped levels. However, the fusion of EDA and BVP signals with ACC data improved DL classifier performance compared to a baseline DL model with ACC-only data. The best performance was achieved by a classifier trained on a combination of ACC, EDA, and BVP images, yielding F1-scores of 69 and 87 for individual and grouped activity classifications, respectively. For DL models trained with additional biological signals, almost all individual activity classifications showed improvement (p-value < 0.05). In grouped activity classifications, DL model performance was enhanced for low- and medium-intensity activities. Exploring the classification of two specific activities, ascending/descending stairs and cycling, revealed significantly improved results using a DL model trained on combined ACC, BVP, and EDA spectrogram images (p-value < 0.05).

The intensity gradient is a new cutpoint-free metric that was developed to quantify physical activity (PA) measured using accelerometers. This metric was developed for use with the ENMO (Euclidean norm minus one) metric, derived from raw acceleration data, and has not been validated for use with count-based accelerometer data. In this study, we determined whether the intensity gradient could be reproduced using count-based accelerometer data. Twenty participants (aged 7–22 years) wore a GT1M, an ActiGraph (count-based), and a GT9X, ActiGraph (raw accelerations) accelerometer during both in-lab and at-home protocols. We found strong agreement between GT1M and GT9X counts during the combined in-lab activities (mean bias = 2 counts) and between minutes per day with different intensities of activity (e.g., sedentary, light, moderate, and vigorous) classified using cutpoints (mean bias < 5 min/d at all intensities). We generated bin sizes that could be used to generate IGs from the count data (mean bias = −0.15; 95% LOA [−0.65, 0.34]) compared with the original IG. Therefore, the intensity gradient could be used to analyze count data. The count-based intensity gradient metric will be valuable for re-analyzing historical datasets collected using older accelerometer models, such as the GT1M.

Participant movement is a major source of artifacts in functional near-infrared spectroscopy (fNIRS) experiments. Mitigating the impact of motion artifacts (MAs) is crucial to estimate brain activity robustly. Here, we suggest and evaluate a novel application of the nonlinear Hammerstein–Wiener model to estimate and mitigate MAs in fNIRS signals from direct-movement recordings through IMU sensors mounted on the participant’s head (head-IMU) and the fNIRS probe (probe-IMU). To this end, we analyzed the hemodynamic responses of single-channel oxyhemoglobin (HbO) and deoxyhemoglobin (HbR) signals from 17 participants who performed a hand tapping task with different levels of concurrent head movement. Additionally, the tapping task was performed without head movements to estimate the ground-truth brain activation. We compared the performance of our novel approach with the probe-IMU and head-IMU to eight established methods (PCA, tPCA, spline, spline Savitzky–Golay, wavelet, CBSI, RLOESS, and WCBSI) on four quality metrics: SNR, △AUC, RMSE, and R. Our proposed nonlinear Hammerstein–Wiener method achieved the best SNR increase (p < 0.001) among all methods. Visual inspection revealed that our approach mitigated MA contaminations that other techniques could not remove effectively. MA correction quality was comparable with head- and probe-IMUs.

Shoe-based wearable sensor systems are a growing research area in health monitoring, disease diagnosis, rehabilitation, and sports training. These systems—equipped with one or more sensors, either of the same or different types—capture information related to foot movement or pressure maps beneath the foot. This captured information offers an overview of the subject’s overall movement, known as the human gait. Beyond sensing, these systems also provide a platform for hosting ambient energy harvesters. They hold the potential to harvest energy from foot movements and operate related low-power devices sustainably. This article proposes two types of strategies (Strategy 1 and Strategy 2) for an energy-autonomous shoe-based system. Strategy 1 uses an accelerometer as a sensor for gait acquisition, which reflects the classical choice. Strategy 2 uses a piezoelectric element for the same, which opens up a new perspective in its implementation. In both strategies, the piezoelectric elements are used to harvest energy from foot activities and operate the system. The article presents a fair comparison between both strategies in terms of power consumption, accuracy, and the extent to which piezoelectric energy harvesters can contribute to overall power management. Moreover, Strategy 2, which uses piezoelectric elements for simultaneous sensing and energy harvesting, is a power-optimized method for an energy-autonomous shoe system.

Kumite is a karate sparring competition in which two players face off and perform offensive and defensive techniques. Depending on the players, there may be preliminary actions (hereinafter referred to as “pre-actions”), such as pulling the arms or legs, lowering the shoulders, etc., just before a technique is performed. Since the presence of a pre-action allows the opponent to know the timing of the technique, it is important to reduce pre-actions in order to improve the kumite. However, it is difficult for beginners and intermediate players to accurately identify their pre-actions and to improve them through practice. Therefore, this study aims to construct a practice support system that enables beginners and intermediate players to understand their pre-actions. In this paper, we focus on the forefist punch, one of kumite’s punching techniques. We propose a method to estimate the presence or absence of a pre-action based on the similarity between the acceleration data of an arbitrary forefist punch and a previously prepared dataset consisting of acceleration data of the forefist punch without a pre-action. We found that the proposed method can estimate the presence or absence of a pre-action in an arbitrary forefist punch with an accuracy of 86%. We also developed KARATECH as a system to support the practice of reducing pre-actions using the proposed method. KARATECH shows the presence or absence of pre-actions through videos and graphs. The evaluation results confirmed that the group using KARATECH had a lower pre-action rate.

Motor fluctuations in Parkinson's disease are characterized by unpredictability in the timing and duration of dopaminergic therapeutic benefits on symptoms, including bradykinesia and rigidity. These fluctuations significantly impair the quality of life of many Parkinson's patients. However, current clinical evaluation tools are not designed for the continuous, naturalistic (real-world) symptom monitoring needed to optimize clinical therapy to treat fluctuations. Although commercially available wearable motor monitoring, used over multiple days, can augment neurological decision making, the feasibility of rapid and dynamic detection of motor fluctuations is unclear. So far, applied wearable monitoring algorithms are trained on group data. In this study, we investigated the influence of individual model training on short timescale classification of naturalistic bradykinesia fluctuations in Parkinson's patients using a single-wrist accelerometer. As part of the Parkinson@Home study protocol, 20 Parkinson patients were recorded with bilateral wrist accelerometers for a one hour OFF medication session and a one hour ON medication session during unconstrained activities in their own homes. Kinematic metrics were extracted from the accelerometer data from the bodyside with the largest unilateral bradykinesia fluctuations across medication states. The kinematic accelerometer features were compared over the 1 h duration of recording, and medication-state classification analyses were performed on 1 min segments of data. Then, we analyzed the influence of individual versus group model training, data window length, and total number of training patients included in group model training, on classification. Statistically significant areas under the curves (AUCs) for medication induced bradykinesia fluctuation classification were seen in 85% of the Parkinson patients at the single minute timescale using the group models. Individually trained models performed at the same level as the group trained models (mean AUC both 0.70, standard deviation respectively 0.18 and 0.10) despite the small individual training dataset. AUCs of the group models improved as the length of the feature windows was increased to 300 s, and with additional training patient datasets. We were able to show that medication-induced fluctuations in bradykinesia can be classified using wrist-worn accelerometry at the time scale of a single minute. Rapid, naturalistic Parkinson motor monitoring has the clinical potential to evaluate dynamic symptomatic and therapeutic fluctuations and help tailor treatments on a fast timescale.

Hip-worn accelerometers are commonly used to assess habitual physical activity, but their accuracy in precisely measuring sedentary behavior (SB) is generally considered low. The angle for postural estimation (APE) method has shown promising accuracy in SB measurement. This method relies on the constant nature of Earth’s gravity and the assumption that walking posture is typically upright. This study investigated how cardiorespiratory fitness (CRF) and body mass index (BMI) are related to APE output. A total of 3475 participants with adequate accelerometer wear time were categorized into three groups according to CRF or BMI. Participants in low CRF and high BMI groups spent more time in reclining and lying postures (APE ≥ 30°) and less time in sitting and standing postures (APE < 30°) than the other groups. Furthermore, the strongest partial Spearman correlation with CRF (r = 0.284) and BMI (r = −0.320) was observed for APE values typical for standing. The findings underscore the utility of the APE method in studying associations between SB and health outcomes. Importantly, this study emphasizes the necessity of reserving the term “sedentary behavior” for studies wherein the classification of SB is based on both intensity and posture.

An observational, repeated measures design was used in this study to assess the validity of the Ergotex Inertial Measurement Unit (IMU) against a 3D motion capture system for measuring trunk, hip, and shoulder angles in ten healthy adult males (38.8 ± 7.3 y, bodyweight 79.2 ± 115.9 kg, body height 179.1 ± 8.1 cm). There were minimal systematic differences between the devices, with the most significant discrepancy being 1.4 degrees for the 80-degree target angle, denoting Ergotex’s precision in joint angle measurements. These results were statistically significant (p < 0.001), with predominantly trivial to small effect sizes, indicating high accuracy for clinical and biomechanical applications. Bland–Altman analysis showed Limits of Agreement (LoA) approximately ±2.5 degrees across all angles and positions, with overall LoA ranging from 3.6 to −2.4 degrees, reflecting Ergotex’s consistent performance. Regression analysis indicated uniform variance across measurements, with minor heteroscedastic errors producing a negligible underestimation trend of around 0.5 degrees in some instances. In conclusion, the Ergotex IMU is a reliable tool for accurate joint angle measurements. It offers a practical and cost-effective alternative to more complex systems, particularly in settings where precise measurement is essential.

To date, clinical expert opinion is the gold standard diagnostic technique for Parkinson’s disease (PD), and continuous monitoring is a promising candidate marker. This study assesses the feasibility and performance of a new wearable tool for supporting the diagnosis of Parkinsonian motor syndromes. The proposed method is based on the use of a wrist-worn measuring system, the execution of a passive, continuous recording session, and a computation of two digital biomarkers (i.e., motor activity and rest tremor index). Based on the execution of some motor tests, a second step is provided for the confirmation of the results of passive recording. In this study, fifty-nine early PD patients and forty-one healthy controls were recruited. The results of this study show that: (a) motor activity was higher in controls than in PD with slight tremors at rest and did not significantly differ between controls and PD with mild-to-moderate tremor rest; (b) the tremor index was smaller in controls than in PD with mild-to-moderate tremor rest and did not significantly differ between controls and PD patients with slight tremor rest; (c) the combination of the said two motor parameters improved the performances in differentiating controls from PD. These preliminary findings demonstrate that the combination of said two digital biomarkers allowed us to differentiate controls from early PD.

Various accelerometry protocols have been used to quantify upper extremity (UE) activity, encompassing diverse epoch lengths and thresholding methods. However, there is no consensus on the most effective approach. The aim of this study was to delineate the optimal parameters for analyzing accelerometry data to quantify UE use in individuals with unilateral cerebral palsy (CP). Methods: A group of adults with CP (n = 15) participated in six activities of daily living, while a group of children with CP (n = 14) underwent the Assisting Hand Assessment. Both groups performed the activities while wearing ActiGraph GT9X-BT devices on each wrist, with concurrent video recording. Use ratio (UR) derived from accelerometry and video analysis and accelerometer data were compared for different epoch lengths (1, 1.5, and 2 s) and activity count (AC) thresholds (between 2 and 150). Results: In adults, results are comparable across epoch lengths, with the best AC thresholds being ≥ 100. In children, results are similar across epoch lengths of 1 and 1.5 (optimal AC threshold = 50), while the optimal threshold is higher with an epoch length of 2 (AC = 75). Conclusions: The combination of epoch length and AC thresholds should be chosen carefully as both influence the validity of the quantification of UE use.

Estimation of temporospatial clinical features of gait (CFs), such as step count and length, step duration, step frequency, gait speed, and distance traveled, is an important component of community-based mobility evaluation using wearable accelerometers. However, accurate unsupervised computerized measurement of CFs of individuals with Duchenne muscular dystrophy (DMD) who have progressive loss of ambulatory mobility is difficult due to differences in patterns and magnitudes of acceleration across their range of attainable gait velocities. This paper proposes a novel calibration method. It aims to detect steps, estimate stride lengths, and determine travel distance. The approach involves a combination of clinical observation, machine-learning-based step detection, and regression-based stride length prediction. The method demonstrates high accuracy in children with DMD and typically developing controls (TDs) regardless of the participant’s level of ability. Fifteen children with DMD and fifteen TDs underwent supervised clinical testing across a range of gait speeds using 10 m or 25 m run/walk (10 MRW, 25 MRW), 100 m run/walk (100 MRW), 6-min walk (6 MWT), and free-walk (FW) evaluations while wearing a mobile-phone-based accelerometer at the waist near the body’s center of mass. Following calibration by a trained clinical evaluator, CFs were extracted from the accelerometer data using a multi-step machine-learning-based process and the results were compared to ground-truth observation data. Model predictions vs. observed values for step counts, distance traveled, and step length showed a strong correlation (Pearson’s r = −0.9929 to 0.9986, p < 0.0001). The estimates demonstrated a mean (SD) percentage error of 1.49% (7.04%) for step counts, 1.18% (9.91%) for distance traveled, and 0.37% (7.52%) for step length compared to ground-truth observations for the combined 6 MWT, 100 MRW, and FW tasks. Our study findings indicate that a single waist-worn accelerometer calibrated to an individual’s stride characteristics using our methods accurately measures CFs and estimates travel distances across a common range of gait speeds in both DMD and TD peers.

This study investigates the feasibility and functionality of accelerometer and gyroscope sensors for gesture-based interactions in mobile app user experience. The core of this innovative approach lies in introducing a dynamic and intuitive user interaction model with the device sensors. The Android app developed for this purpose has been created for its use in controlled experiments. Methodologically, it was created as a stand-alone tool to both capture quantitative (time, automatically captured) and qualitative (behavior, collected with post-task questionnaires) variables. The app’s setting features a set of modules with two levels each (randomized presentation applied, minimizing potential learning effects), allowing users to interact with both sensor-based and traditional touch-based scenarios. Preliminary results with 22 participants reveal that tasks involving sensor-based interactions tend to take longer to complete when compared to the traditional ones. Remarkably, many participants rated sensor-based interactions as a better option than touch-based interactions, as seen in the post-task questionnaires. This apparent discrepancy between objective completion times and subjective user perceptions requires a future in-depth exploration of factors influencing user experiences, including potential learning curves, cognitive load, and task complexity. This study contributes to the evolving landscape of mobile app user experience, emphasizing the benefits of considering the integration of device sensors (and gesture-based interactions) in common mobile usage.

The literature has yielded promising data over the past decade regarding the use of inertial sensors for the analysis of occupational ergonomics. However, despite their significant advantages (e.g., portability, lightness, low cost, etc.), their widespread implementation in the actual workplace has not yet been realized, possibly due to their discomfort or potential alteration of the worker’s behaviour. This systematic review has two main objectives: (i) to synthesize and evaluate studies that have employed inertial sensors in ergonomic analysis based on the RULA method; and (ii) to propose an evaluation system for the transparency of this technology to the user as a potential factor that could influence the behaviour and/or movements of the worker. A search was conducted on the Web of Science and Scopus databases. The studies were summarized and categorized based on the type of industry, objective, type and number of sensors used, body parts analysed, combination (or not) with other technologies, real or controlled environment, and transparency. A total of 17 studies were included in this review. The Xsens MVN system was the most widely used in this review, and the majority of studies were classified with a moderate level of transparency. It is noteworthy, however, that there is a limited and worrisome number of studies conducted in uncontrolled real environments.

Background: Self-reported adherence to sling wear is unreliable due to recall bias. We aim to assess the feasibility and accuracy of quantifying sling wear and non-wear utilising slings pre-fitted with a GENEActiv accelerometer that houses triaxial acceleration and temperature sensors. Methods: Ten participants were asked to wear slings for 480 min (8 h) incorporating 180 min of non-wear time in durations varying from 5–120 min. GENEActiv devices were fitted in sutured inner sling pockets and participants logged sling donning and doffing times. An algorithm based on variability in acceleration in three axes and temperature change was developed to identify sling wear and non-wear and compared to participants’ logs. Results: There was no significant difference between algorithm detected non-wear duration (mean ± standard deviation = 172.0 ± 6.8 min/participant) and actual non-wear (179.7 ± 1.0 min/participant). Minute-by-minute agreement of sensor-detected wear and non-wear with participant reported wear was 97.3 ± 1.5% (range = 93.9–99.0), with mean sensitivity 94.3 ± 3.5% (range = 86.1–98.3) and specificity 99.1 ± 0.8% (range = 93.7–100). Conclusion: An algorithm based on accelerometer-assessed acceleration and temperature can accurately identify shoulder sling wear/non-wear times. This method may have potential for assessing whether sling wear adherence after shoulder surgeries have any bearing on patient functional outcomes.

The problem of remaining useful life estimation (RULE) of hollow worn railway vehicle wheels in terms of remaining mileage via wheel tread depth estimation using on-board vibration signals from a single accelerometer on the bogie frame is presently investigated. This is achieved based on the introduction of a statistical time series method that employs: (i) advanced data-driven stochastic Functionally Pooled models for the modeling of the vehicle dynamics under different wheel tread depths in a range of interest until a critical limit, as well as tread depth estimation through a proper optimization procedure, and (ii) a wheel tread depth evolution function with respect to the vehicle running mileage that interconnects the estimated hollow wear with the remaining useful mileage. The method’s RULE performance is investigated via hundreds of Simpack-based Monte Carlo simulations with an Attiko Metro S.A. vehicle and many hollow worn wheels scenarios which are not used for the method’s training. The obtained results indicate the accurate estimation of the wheels tread depth with a mean absolute error of ∼0.07 mm that leads to a corresponding small error of ∼3% with respect to the wheels remaining useful mileage. In addition, the comparison with a recently introduced Multiple Model (MM)-based multi-health state classification method for RULE, demonstrates the better performance of the postulated method that achieves 81.17% True Positive Rate (TPR) which is significantly higher than the 45.44% of the MM method.

Many stroke survivors experience persistent upper extremity impairment that limits performance in activities of daily living. Upper limb recovery requires high repetitions of task-specific practice. Stroke survivors are often prescribed task practices at home to supplement rehabilitation therapy. A poor quality of task practices, such as the use of compensatory movement patterns, results in maladaptive neuroplasticity and suboptimal motor recovery. There currently lacks a tool for the remote monitoring of movement quality of stroke survivors’ task practices at home. The objective of this study was to evaluate the feasibility of classifying movement quality at home using a wearable IMU. Nineteen stroke survivors wore an IMU sensor on the paretic wrist and performed four functional upper limb tasks in the lab and later at home while videorecording themselves. The lab data served as reference data to classify home movement quality using dynamic time warping. Incorrect and correct movement quality was labeled by a therapist. The home task practice movement quality was classified with an accuracy of 92% and F1 score of 0.95 for all tasks combined. Movement types contributing to misclassification were further investigated. The results support the feasibility of a home movement quality monitoring system to assist with upper limb rehabilitation post stroke.

In overcoming the worldwide problem of overweight and obesity, automatic dietary monitoring (ADM) is introduced as support in dieting practises. ADM aims to automatically, continuously, and objectively measure dimensions of food intake in a free-living environment. This could simplify the food registration process, thereby overcoming frequent memory, underestimation, and overestimation problems. In this study, an eating event detection sensor system was developed comprising a smartwatch worn on the wrist containing an accelerometer and gyroscope for eating gesture detection, a piezoelectric sensor worn on the jaw for chewing detection, and a respiratory inductance plethysmographic sensor consisting of two belts worn around the chest and abdomen for food swallowing detection. These sensors were combined to determine to what extent a combination of sensors focusing on different steps of the dietary cycle can improve eating event classification results. Six subjects participated in an experiment in a controlled setting consisting of both eating and non-eating events. Features were computed for each sensing measure to train a support vector machine model. This resulted in F1-scores of 0.82 for eating gestures, 0.94 for chewing food, and 0.58 for swallowing food.

Fall accidents in the construction industry have been studied over several decades and identified as a common hazard and the leading cause of fatalities. Inertial sensors have recently been used to detect accidents of workers in construction sites, such as falls or trips. IMU-based systems for detecting fall-related accidents have been developed and have yielded satisfactory accuracy in laboratory settings. Nevertheless, the existing systems fail to uphold consistent accuracy and produce a significant number of false alarms when deployed in real-world settings, primarily due to the intricate nature of the working environments and the behaviors of the workers. In this research, the authors redesign the aforementioned laboratory experiment to target situations that are prone to false alarms based on the feedback obtained from workers in real construction sites. In addition, a new algorithm based on recurrent neural networks was developed to reduce the frequencies of various types of false alarms. The proposed model outperforms the existing benchmark model (i.e., hierarchical threshold model) with higher sensitivities and fewer false alarms in detecting stumble (100% sensitivity vs. 40%) and fall (95% sensitivity vs. 65%) events. However, the model did not outperform the hierarchical model in detecting coma events in terms of sensitivity (70% vs. 100%), but it did generate fewer false alarms (5 false alarms vs. 13).

Smartphone applications (apps) that utilize embedded inertial sensors have the potential to provide valid and reliable estimations of different balance and gait parameters in older adults with mild balance impairment. This study aimed to assess the reliability, validity, and sensitivity of the Gait&Balance smartphone application (G&B App) for measuring gait and balance in a sample of middle- to older-aged adults with mild balance impairment in Pakistan. Community-dwelling adults over 50 years of age (N = 83, 50 female, range 50–75 years) with a Berg Balance Scale (BBS) score between 46/56 and 54/56 were included in the study. Data collection involved securing a smartphone to the participant’s lumbosacral spine. Participants performed six standardized balance tasks, including four quiet stance tasks and two gait tasks (walking looking straight ahead and walking with head turns). The G&B App collected accelerometry data during these tasks, and the tasks were repeated twice to assess test-retest reliability. The tasks in quiet stance were also recorded with a force plate, a gold-standard technology for measuring postural sway. Additionally, participants completed three clinical measures, the BBS, the Functional Reach Test (FRT), and the Timed Up and Go Test (TUG). Test-retest reliability within the same session was determined using intraclass correlation coefficients (ICCs) and the standard error of measurement (SEM). Validity was evaluated by correlating the G&B App outcomes against both the force plate data and the clinical measures using Pearson’s product-moment correlation coefficients. To assess the G&B App’s sensitivity to differences in balance across tasks and repetitions, one-way repeated measures analyses of variance (ANOVAs) were conducted. During quiet stance, the app demonstrated moderate reliability for steadiness on firm (ICC = 0.72) and compliant surfaces (ICC = 0.75) with eyes closed. For gait tasks, the G&B App indicated moderate to excellent reliability when walking looking straight ahead for gait symmetry (ICC = 0.65), walking speed (ICC = 0.93), step length (ICC = 0.94), and step time (ICC = 0.84). The TUG correlated with app measures under both gait conditions for walking speed (r −0.70 and 0.67), step length (r −0.56 and −0.58), and step time (r 0.58 and 0.50). The BBS correlated with app measures of walking speed under both gait conditions (r 0.55 and 0.51) and step length when walking with head turns (r = 0.53). Force plate measures of total distance wandered showed adequate to excellent correlations with G&B App measures of steadiness. Notably, G&B App measures of walking speed, gait symmetry, step length, and step time, were sensitive to detecting differences in performance between standard walking and the more difficult task of walking with head turns. This study demonstrates the G&B App’s potential as a reliable and valid tool for assessing some gait and balance parameters in middle-to-older age adults, with promise for application in low-income countries like Pakistan. The app’s accessibility and accuracy could enhance healthcare services and support preventive measures related to fall risk.

Although the vibration of rackets and the location of the sweet spot for players when hitting the ball is crucial, manufacturers do not specify this behavior precisely. This article analyses padel rackets, provides a solution to determine the sweet spot position (SSP), quantifies its behavior, and determines the level of vibration transmitted along the racket handle. The proposed methods serve to locate the SSP without quantifying it. This article demonstrates the development of equipment capable of analyzing the vibration behavior of padel rackets. To do so, it employs a robot that moves along the surface of the padel racket, striking it along its central line. Accelerometers are placed on a movable cradle where rackets are positioned and adjusted. A method for analyzing accelerometer signals to quantify vibration severity is proposed. The SSP and vibration behavior along the central line are determined and quantified. As a result of the study, 225 padel rackets are analyzed and compared. SSP is independent of the padel racket shape, balance, weight, moment of inertia, and padel racket shape (tear, diamond, or round) and is not located at the same position as the center of percussion.

Estimation of oxygen consumption (VO2) from accelerometer data is typically based on prediction equations developed in laboratory settings using steadily paced and controlled test activities. These equations may not capture the temporary changes in VO2 occurring in sporadic real-life physical activity. In this study, we introduced a novel floating epoch for accelerometer data analysis and hypothesized that an adaptive epoch length provides a more consistent estimation of VO2 in irregular activity conditions than a 6 s constant epoch. Two different activity tests were conducted: a progressive constant-speed test (CS) performed on a track and a 6 min back-and-forth walk test including accelerations and decelerations (AC/DC) performed as fast as possible. Twenty-nine adults performed the CS test, and sixty-one performed the AC/DC test. The data were collected using hip-worn accelerometers and a portable metabolic gas analyzer. General linear models were employed to create the prediction models for VO2 that were cross-validated using both data sets and epoch types as training and validation sets. The prediction equations based on the CS test or AC/DC test and 6 s epoch had excellent performance (R2 = 89%) for the CS test but poor performance for the AC/DC test (31%). Only the VO2 prediction equation based on the AC/DC test and the floating epoch had good performance (78%) for both tests. The overall accuracy of VO2 prediction is compromised with the constant length epoch, whereas the prediction model based on irregular acceleration data analyzed with a floating epoch provided consistent performance for both activities.

The main object of this research was to develop a security system to evaluate the intrusion into an object through a glass pane. More specifically, this study deals with sensing and evaluating signals from a contact glass-break detector, which is part of an intruder alarm system. Each alarm detector in an alarm system must accomplish certain security level requirements that strictly describe the requirements for the area of use and the detector’s reliability. To date, no contact glass-break detector has been developed and fully tested to meet the stringent requirements of the highest security level. A contact glass-break detector was developed whose main part is an accelerometer that transmits signals from the glass pane. These signals were evaluated according to the developed methodology. It was verified that the proposed system can distinguish at the highest security level between false alarms and situations where the building has been intruded.

Accurate and reliable measurement of real-world walking activity is clinically relevant, particularly for people with mobility difficulties. Insights on walking can help understand mobility function, disease progression, and fall risks. People living in long-term residential care environments have heterogeneous and often pathological walking patterns, making it difficult for conventional algorithms paired with wearable sensors to detect their walking activity. We designed two walking bout detection algorithms for people living in long-term residential care. Both algorithms used thresholds on the magnitude of acceleration from a 3-axis accelerometer on the lower back to classify data as “walking” or “non-walking”. One algorithm had generic thresholds, whereas the other used personalized thresholds. To validate and evaluate the algorithms, we compared the classifications of walking/non-walking from our algorithms to the real-time research assistant annotated labels and the classification output from an algorithm validated on a healthy population. Both the generic and personalized algorithms had acceptable accuracy (0.83 and 0.82, respectively). The personalized algorithm showed the highest specificity (0.84) of all tested algorithms, meaning it was the best suited to determine input data for gait characteristic extraction. The developed algorithms were almost 60% quicker than the previously developed algorithms, suggesting they are adaptable for real-time processing.

With the expansion of electronics in recent decades, it is notorious to observe that embedded systems are increasingly necessary to improve people’s quality of life and to facilitate the diagnosis of systems in general, ranging from pacemakers to control systems. The increased use of electronic components for technological support, such as telemetry systems, electronic injection, and automotive diagnostic scanners, enhances the perspective of data analysis through an embedded system aimed at vehicular systems. Thus, this work aims to design and implement an embedded data acquisition system for the analysis of vehicle vertical dynamics. The methodology for this study was structured into several stages: mathematical modeling of a motorcycle’s mass-spring-damper system, coding for the Arduino microcontroller, computational data analysis supported by MATLAB software version 9.6, electronic prototyping of the embedded system, implementation on the vehicle, and the analysis of motorcycle vertical dynamics parameters. In addition, a mathematical modeling of the mass-spring-damper system was performed using the state-space method. The system was implemented on the Arduino microcontroller platform, enabling real-time data transfer from a motorcycle. The experimental results have successfully validated the proposed data acquisition system.

Electric scooters are quickly becoming a popular form of mobility in many cities around the world, which has led to a surge in safety incidents. Moreover, electric scooters are not equipped with safety devices for riders, which can lead to serious accidents. In this study, a footrest, data-collection module, and accelerometer module for electric scooters were developed to prevent various accidents caused by the rapid increase in the use of electric scooters. In the experiment, the boarding data of the electric-scooter riders were collected from the footrest and data-collection module. Moreover, the driving data of the electric scooters for different road types were collected with the accelerometer module. We then trained two artificial intelligence (AI) models based on convolutional neural networks (CNNs) for different types of data. When we considered the learning accuracy and mean square error (MSE), which are performance indicators of the ability of the trained AI models to discriminate data, for each AI model, the learning accuracy converged to 100% and the MSE converged to 0. Further, this study is expected to help reduce the accident rate of electric scooters by resolving the causes of frequent accidents involving electric scooters around the world.

The escalating global water usage and the increasing strain on major cities due to water shortages highlights the critical need for efficient water management practices. In water-stressed regions worldwide, significant water wastage is primarily attributed to leakages, inefficient use, and aging infrastructure. Undetected water leakages in buildings’ pipelines contribute to the water waste problem. To address this issue, an effective water leak detection method is required. In this paper, we explore the application of edge computing in smart buildings to enhance water management. By integrating sensors and embedded Machine Learning models, known as TinyML, smart water management systems can collect real-time data, analyze it, and make accurate decisions for efficient water utilization. The transition to TinyML enables faster and more cost-effective local decision-making, reducing the dependence on centralized entities. In this work, we propose a solution that can be adapted for effective leakage detection in real-world scenarios with minimum human intervention using TinyML. We follow an approach that is similar to a typical machine learning lifecycle in production, spanning stages including data collection, training, hyperparameter tuning, offline evaluation and model optimization for on-device resource efficiency before deployment. In this work, we considered an existing water leakage acoustic dataset for polyvinyl chloride pipelines. To prepare the acoustic data for analysis, we performed preprocessing to transform it into scalograms. We devised a water leak detection method by applying transfer learning to five distinct Convolutional Neural Network (CNN) variants, which are namely EfficientNet, ResNet, AlexNet, MobileNet V1, and MobileNet V2. The CNN models were found to be able to detect leakages where a maximum testing accuracy, recall, precision, and F1 score of 97.45%, 98.57%, 96.70%, and 97.63%, respectively, were observed using the EfficientNet model. To enable seamless deployment on the Arduino Nano 33 BLE edge device, the EfficientNet model is compressed using quantization resulting in a low inference time of 1932 ms, a peak RAM usage of 255.3 kilobytes, and a flash usage requirement of merely 48.7 kilobytes.

Real-world gait analysis can aid in clinical assessments and influence related interventions, free from the restrictions of a laboratory setting. Using individual accelerometers, we aimed to use a simple machine learning method to quantify the performance of the discrimination between three self-selected cyclical locomotion types using accelerometers placed at frequently referenced attachment locations. Thirty-five participants walked along a 10 m walkway at three different speeds. Triaxial accelerometers were attached to the sacrum, thighs and shanks. Slabs of magnitude, three-second-long accelerometer data were transformed into two-dimensional Fourier spectra. Principal component analysis was undertaken for data reduction and feature selection, followed by discriminant function analysis for classification. Accuracy was quantified by calculating scalar accounting for the distances between the three centroids and the scatter of each category’s cloud. The algorithm could successfully discriminate between gait modalities with 91% accuracy at the sacrum, 90% at the shanks and 87% at the thighs. Modalities were discriminated with high accuracy in all three sensor locations, where the most accurate location was the sacrum. Future research will focus on optimising the data processing of information from sensor locations that are advantageous for practical reasons, e.g., shank for prosthetic and orthotic devices.

Parkinson’s disease (PD) is a neurodegenerative disorder characterized by motor and nonmotor impairment with various implications on patients’ quality of life. Since currently available therapies are only symptomatic, identifying individuals with prodromal, preclinical, or early-stage PD is crucial, as they would be ideal candidates for future disease-modifying therapies. Our analysis aims to develop a robust model for accurate PD detection using accelerometer data collected from PD and non-PD individuals with mild or no tremor during phone conversations. An open-access dataset comprising accelerometer recordings from 22 PD patients and 11 healthy controls (HCs) was utilized. The data were preprocessed to extract relevant time-, frequency-, and energy-related features, and a bidirectional long short-term memory (Bi-LSTM) model with attention mechanism was employed for classification. The performance of the model was evaluated using fivefold cross-validation, and metrics of accuracy, precision, recall, specificity, and f1-score were computed. The proposed model demonstrated high accuracy (98%), precision (99%), recall (98%), specificity (96%), and f1-score (98%) in accurately distinguishing PD patients from HCs. Our findings indicate that the proposed model outperforms existing approaches and holds promise for detection of PD with subtle symptoms, like tremor, in the wild. Such symptoms can present in the early or even prodromal stage of the disease, and appropriate mobile-based applications may be a practical tool in real-life settings to alert individuals at risk to seek medical assistance or give patients feedback in monitoring their symptoms.

Wrist-based respiratory rate (RR) measurement during sleep faces accuracy limitations. This study aimed to assess the accuracy of the RR estimation function during sleep based on the severity of obstructive sleep apnea (OSA) using the Samsung Galaxy Watch (GW) series. These watches are equipped with accelerometers and photoplethysmography sensors for RR estimation. A total of 195 participants visiting our sleep clinic underwent overnight polysomnography while wearing the GW, and the RR estimated by the GW was compared with the reference RR obtained from the nasal thermocouple. For all participants, the root mean squared error (RMSE) of the average overnight RR and continuous RR measurements were 1.13 bpm and 1.62 bpm, respectively, showing a small bias of 0.39 bpm and 0.37 bpm, respectively. The Bland–Altman plots indicated good agreement in the RR measurements for the normal, mild, and moderate OSA groups. In participants with normal-to-moderate OSA, both average overnight RR and continuous RR measurements achieved accuracy rates exceeding 90%. However, for patients with severe OSA, these accuracy rates decreased to 79.45% and 75.8%, respectively. The study demonstrates the GW’s ability to accurately estimate RR during sleep, even though accuracy may be compromised in patients with severe OSA.

Intelligent compaction (IC) is a technology that uses non-contact sensors to monitor and record the compaction level of geomaterials in real-time during road construction. However, current IC devices have several limitations: (i) they are unable to visualize or compare multiple intelligent compaction measurement values (ICMVs) in real-time during compaction; (ii) they are not retrofittable to different conventional rollers that exist in the field; (iii) they do not incorporate corrections for ICMVs reflecting variable field conditions; (iv) they are unable to integrate construction specifications as needed for performance-based compaction; and (v) they do not record all the key roller parameters for further compaction analysis. To address these issues, an innovative retrofittable platform with cutting-edge hardware and software was developed. This platform, called the intelligent compaction analyzer (ICA) platform, is effective at calculating conventional acceleration amplitude-based ICMVs and stiffness-based parameters and at displaying the spatial distributions of these parameters in a color-coded map in real-time during compaction.

A three-dimensional motion capture system (MoCap) and the Garmin Running Dynamics Pod can be utilised to monitor a variety of dynamic parameters during running. The present investigation was designed to examine the validity of these two systems for determining ground contact times while running in place by comparing the values obtained with those provided by the bilateral force plate (gold standard). Eleven subjects completed three 20-s runs in place at self-selected rates, starting slowly, continuing at an intermediate pace, and finishing rapidly. The ground contact times obtained with both systems differed significantly from the gold standard at all three rates, as well as for all the rates combined (p < 0.001 in all cases), with the smallest mean bias at the fastest step rate for both (11.5 ± 14.4 ms for MoCap and −81.5 ± 18.4 ms for Garmin). This algorithm was developed for the determination of ground contact times during normal running and was adapted here for the assessment of running in place by the MoCap, which could be one explanation for its lack of validity. In conclusion, the wearables developed for monitoring normal running cannot be assumed to be suitable for determining ground contact times while running in place.

Animal behaviour can be an indicator of health and welfare. Monitoring behaviour through visual observation is labour-intensive and there is a risk of missing infrequent behaviours. Twelve healthy domestic shorthair cats were fitted with triaxial accelerometers mounted on a collar and harness. Over seven days, accelerometer and video footage were collected simultaneously. Identifier variables (n = 32) were calculated from the accelerometer data and summarized into 1 s epochs. Twenty-four behaviours were annotated from the video recordings and aligned with the summarised accelerometer data. Models were created using random forest (RF) and supervised self-organizing map (SOM) machine learning techniques for each mounting location. Multiple modelling rounds were run to select and merge behaviours based on performance values. All models were then tested on a validation accelerometer dataset from the same twelve cats to identify behaviours. The frequency of behaviours was calculated and compared using Dirichlet regression. Despite the SOM models having higher Kappa (>95%) and overall accuracy (>95%) compared with the RF models (64–76% and 70–86%, respectively), the RF models predicted behaviours more consistently between mounting locations. These results indicate that triaxial accelerometers can identify cat specific behaviours.

The study of Domain Generalization (DG) has gained considerable momentum in the Machine Learning (ML) field. Human Activity Recognition (HAR) inherently encompasses diverse domains (e.g., users, devices, or datasets), rendering it an ideal testbed for exploring Domain Generalization. Building upon recent work, this paper investigates the application of regularization methods to bridge the generalization gap between traditional models based on handcrafted features and deep neural networks. We apply various regularizers, including sparse training, Mixup, Distributionally Robust Optimization (DRO), and Sharpness-Aware Minimization (SAM), to deep learning models and assess their performance in Out-of-Distribution (OOD) settings across multiple domains using homogenized public datasets. Our results show that Mixup and SAM are the best-performing regularizers. However, they are unable to match the performance of models based on handcrafted features. This suggests that while regularization techniques can improve OOD robustness to some extent, handcrafted features remain superior for domain generalization in HAR tasks.

Measures of physical performance captured within a clinical setting are commonly used as a surrogate for underlying health or disease risk within an individual. By measuring physical behaviour within a free-living setting, we may be able to better quantify physical performance. In our study, we outline an approach to measure maximum free-living step count using a body-worn sensor as an indicator of physical performance. We then use this approach to characterise the maximum step count over a range of window durations within a population of older adults to identify a preferred duration over which to measure the maximum step count. We found that while almost all individuals (97%) undertook at least one instance of continuous stepping longer than two minutes, a sizeable minority of individuals (31%) had no periods of continuous stepping longer than six minutes. We suggest that the maximum step count measured over a six-minute period may be too sensitive to the adults’ lack of opportunity to undertake prolonged periods of stepping, and a two-minute window could provide a more representative measure of physical performance.

Upper extremity hemiplegia is a serious problem affecting the lives of many people post-stroke. Motor recovery requires high repetitions and quality of task-specific practice. Sufficient practice cannot be completed during therapy sessions, requiring patients to perform additional task practices at home on their own. Adherence to and quality of these home task practices are often limited, which is likely a factor reducing rehabilitation effectiveness post-stroke. However, home adherence is typically measured by self-reports that are known to be inconsistent with objective measurement. The objective of this study was to develop algorithms to enable the objective identification of task type and quality. Twenty neurotypical participants wore an IMU sensor on the wrist and performed four representative tasks in prescribed fashions that mimicked correct, compensatory, and incomplete movement qualities typically seen in stroke survivors. LSTM classifiers were trained to identify the task being performed and its movement quality. Our models achieved an accuracy of 90.8% for task identification and 84.9%, 81.1%, 58.4%, and 73.2% for movement quality classification for the four tasks for unseen participants. The results warrant further investigation to determine the classification performance for stroke survivors and if quantity and quality feedback from objective monitoring facilitates effective task practice at home, thereby improving motor recovery.

Onboard electrostatic suspension inertial sensors are important applications for gravity satellites and space gravitational-wave detection missions, and it is important to suppress noise in the measurement signal. Due to the complex coupling between the working space environment and the satellite platform, the process of noise generation is extremely complex, and traditional noise modeling and subtraction methods have certain limitations. With the development of deep learning, applying it to high-precision inertial sensors to improve the signal-to-noise ratio is a practically meaningful task. Since there is a single noise sample and unknown true value in the measured data in orbit, odd–even sub-samplers and periodic sub-samplers are designed to process general signals and periodic signals, and adds reconstruction layers consisting of fully connected layers to the model. Experimental analysis and comparison are conducted based on simulation data, GRACE-FO acceleration data, and Taiji-1 acceleration data. The results show that the deep learning method is superior to traditional data smoothing processing solutions.

Background: Health care professionals need a valid tool to assess the physical ability of patients with chronic diseases. We aimed to assess the validity of the results of physical fitness tests estimated by a wrist wearable device in young adults and chronic disease people. Methods: Participants wore a sensor placed on their wrist and performed two physical fitness tests (sit to stand (STS) and time up and go (TUG)). We checked the concordance of sensor-estimated results using Bland–Altman analysis, root-mean-square error, and intraclass coefficient of correlation (ICC). Results: In total, 31 young adults (groups A; median age = 25 ± 5 years) and 14 people with chronic diseases (groups B; median age = 70 ± 15 years) were included. Concordance was high for both STS (ICCA = 0.95, and ICCB = 0.90), and TUG (ICCA = 0.75, ICCB = 0.98). The best estimations were given by the sensor during STS tests in young adults (mean bias = 0.19 ± 2.69; p = 0.12) and chronic disease people (mean bias = −0.14 ± 3.09 s; p = 0.24). The sensor provided the largest estimation errors over 2 s during the TUG test in young adults. Conclusion: This study showed that the results provided by the sensor are consistent with those of the gold standard during STS and TUG in both healthy youth and people with chronic diseases.

Sleep is extremely important for physical and mental health. Although polysomnography is an established approach in sleep analysis, it is quite intrusive and expensive. Consequently, developing a non-invasive and non-intrusive home sleep monitoring system with minimal influence on patients, that can reliably and accurately measure cardiorespiratory parameters, is of great interest. The aim of this study is to validate a non-invasive and unobtrusive cardiorespiratory parameter monitoring system based on an accelerometer sensor. This system includes a special holder to install the system under the bed mattress. The additional aim is to determine the optimum relative system position (in relation to the subject) at which the most accurate and precise values of measured parameters could be achieved. The data were collected from 23 subjects (13 males and 10 females). The obtained ballistocardiogram signal was sequentially processed using a sixth-order Butterworth bandpass filter and a moving average filter. As a result, an average error (compared to reference values) of 2.24 beats per minute for heart rate and 1.52 breaths per minute for respiratory rate was achieved, regardless of the subject’s sleep position. For males and females, the errors were 2.28 bpm and 2.19 bpm for heart rate and 1.41 rpm and 1.30 rpm for respiratory rate. We determined that placing the sensor and system at chest level is the preferred configuration for cardiorespiratory measurement. Further studies of the system’s performance in larger groups of subjects are required, despite the promising results of the current tests in healthy subjects.