BACKGROUND: The long-term monitoring of internal and external training load is crucial for the training effectiveness of athletes. This study aims to quantify the internal and external training loads of collegiate male volleyball players during the competitive season. The internal and external training load variables were analyzed across mesocycles and playing positions.
METHODS: Fourteen participants with age of 20.2 ± 1.3 years, height of 1.81 ± 0.05 m, and body weight of 70.8 ± 5.9 kg were recruited. The data were collected over a 29-week period that was divided into four mesocycles: preparation 1 (P1, weeks 1-7), competition 1 (C1, weeks 8-14, including a 5-day tournament in week 14), preparation 2 (P2, weeks 15-23), and competition 2 (C2, weeks 24-29, including a 6-day tournament in week 29). Each participant wore an inertial measurement unit and reported the rating of perceived exertion in each training session. The internal training load variables included weekly session rating of perceived exertion, acute: chronic workload ratio, and training monotony and strain. The external training load variables included jump count and height and the percentage of jumps exceeding 80% of maximal height.
RESULTS: C2 had the highest average weekly internal training load (3022 ± 849 AU), whereas P2 had the highest average weekly acute: chronic workload ratio (1.46 ± 0.13 AU). The number of weekly jumps in C1 (466.0 ± 176.8) was significantly higher than in other mesocycles. Weekly jump height was significantly higher in C1, P2, and C2. Internal training load was positively correlated with jump count (ρ = 0.477, p < 0.001). Jump count was negatively correlated with jump height (ρ = -0.089, p = 0.006) and the percentage of jumps exceeding 80% of maximal height (ρ = -0.388, p < 0.001). The internal and external training load variables were similar among different playing positions.
CONCLUSIONS: The participants exhibited significantly higher internal training load in C2 and higher jump height after P1. A high jump count was associated with higher internal training load and lower jump height. Excessive jumps may result in fatigue and reduce height.

BACKGROUND: Mixed reality (MR) is helpful in hand training for patients with stroke, allowing them to fully submerge in a virtual space while interacting with real objects. The recognition of individual finger movements is required for MR rehabilitation. This study aimed to assess the effectiveness of updated MR-board 2, adding finger training for patients with stroke.
METHODS: Twenty-one participants with hemiplegic stroke (10 with left hemiplegia and 11 with right hemiplegia; nine female patients; 56.7 ± 14.2 years of age; and onset of stroke 32.7 ± 34.8 months) participated in this study. MR-board 2 comprised a board plate, a depth camera, plastic-shaped objects, a monitor, a palm-worn camera, and seven gamified training programs. All participants performed 20 self-training sessions involving 30-min training using MR-board 2. The outcome measurements for upper extremity function were the Fugl-Meyer assessment (FMA) upper extremity score, repeated number of finger flexion and extension (Repeat-FE), the thumb opposition test (TOT), Box and Block Test score (BBT), Wolf Motor Function Test score (WMFT), and Stroke Impact Scale (SIS). One-way repeated measures analysis of variance and the post hoc test were applied for the measurements. MR-board 2 recorded the fingers\' active range of motion (AROM) and Dunnett\'s test was used for pairwise comparisons.
RESULTS: Except for the FMA-proximal score (p = 0.617) and TOT (p = 0.005), other FMA scores, BBT score, Repeat-FE, WMFT score, and SIS stroke recovery improved significantly (p < 0.001) during MR-board 2 training and were maintained until follow-up. All AROM values of the finger joints changed significantly during training (p < 0.001).
CONCLUSIONS: MR-board 2 self-training, which includes natural interactions between humans and computers using a tangible user interface and real-time tracking of the fingers, improved upper limb function across impairment, activity, and participation. MR-board 2 could be used as a self-training tool for patients with stroke, improving their quality of life.
BACKGROUND: This study was registered with the Clinical Research Information Service (CRIS: KCT0004167).

Multi-segment foot kinematics during shod running are difficult to investigate in clinical settings. Stretch strain sensors can measure foot kinematics; however, whether they can evaluate foot kinematics during shod running or at the midfoot kinematics remains unclear. The aim of this study was to investigate the stretch strain sensor could reveal differences between shod and barefoot conditions and midfoot kinematics during running. Eighteen healthy adults were included in the study. A stretch strain sensor and three-dimensional motion capture system were used to measure foot kinematics during barefoot and shod running with a rearfoot strike pattern. The correlation between the amplitudes of the two signals during barefoot running was investigated, and the similarity between the two signals was evaluated using the cross-correlation coefficient. Statistical parametric mapping was used to compare shod and barefoot conditions. Shod running had significantly lower sensor strain from 30 % to 100 % stance compared to barefoot running (p < 0.05). The sensor amplitude was significantly correlated with the shank-rearfoot frontal (r = 0.668, p = 0.002), the rearfoot-midfoot transverse (r = 0.546, p = 0.02), and the midfoot-forefoot sagittal planes (r = 0.563, p = 0.01). A high cross-correlation was observed between the sensor signal and the shank-rearfoot sagittal, frontal, and transverse planes and the midfoot-forefoot sagittal plane. This sensor can be used to investigate foot kinematics during shod running. The sensor signal mainly reflects the shank-rearfoot frontal and midfoot-forefoot sagittal planes, as well as the maximum kinematic range of the rearfoot-midfoot transverse plane.

BACKGROUND: Work characteristics, such as teleworking rate, have been studied in relation to stress. However, the use of work-related data to improve a high-performance stress prediction model that suits an individual\'s lifestyle has not been evaluated.
OBJECTIVE: This study aims to develop a novel, high-performance algorithm to predict an employee\'s stress among a group of employees with similar working characteristics.
METHODS: This prospective observational study evaluated participants\' responses to web‑based questionnaires, including attendance records and data collected using a wearable device. Data spanning 12 weeks (between January 17, 2022, and April 10, 2022) were collected from 194 Shionogi Group employees. Participants wore the Fitbit Charge 4 wearable device, which collected data on daily sleep, activity, and heart rate. Daily work shift data included details of working hours. Weekly questionnaire responses included the K6 questionnaire for depression/anxiety, a behavioral questionnaire, and the number of days lunch was missed. The proposed prediction model used a neighborhood cluster (N=20) with working-style characteristics similar to those of the prediction target person. Data from the previous week predicted stress levels the following week. Three models were compared by selecting appropriate training data: (1) single model, (2) proposed method 1, and (3) proposed method 2. Shapley Additive Explanations (SHAP) were calculated for the top 10 extracted features from the Extreme Gradient Boosting (XGBoost) model to evaluate the amount and contribution direction categorized by teleworking rates (mean): low: <0.2 (more than 4 days/week in office), middle: 0.2 to <0.6 (2 to 4 days/week in office), and high: ≥0.6 (less than 2 days/week in office).
RESULTS: Data from 190 participants were used, with a teleworking rate ranging from 0% to 79%. The area under the curve (AUC) of the proposed method 2 was 0.84 (true positive vs false positive: 0.77 vs 0.26). Among participants with low teleworking rates, most features extracted were related to sleep, followed by activity and work. Among participants with high teleworking rates, most features were related to activity, followed by sleep and work. SHAP analysis showed that for participants with high teleworking rates, skipping lunch, working more/less than scheduled, higher fluctuations in heart rate, and lower mean sleep duration contributed to stress. In participants with low teleworking rates, coming too early or late to work (before/after 9 AM), a higher/lower than mean heart rate, lower fluctuations in heart rate, and burning more/fewer calories than normal contributed to stress.
CONCLUSIONS: Forming a neighborhood cluster with similar working styles based on teleworking rates and using it as training data improved the prediction performance. The validity of the neighborhood cluster approach is indicated by differences in the contributing features and their contribution directions among teleworking levels.
BACKGROUND: UMIN UMIN000046394; https://www.umin.ac.jp/ctr/index.htm.

A two-stream convolutional neural network (TCNN) for breathing pattern classification has been devised for the continuous monitoring of patients with infectious respiratory diseases. The TCNN consists of a convolutional neural network (CNN)-based autoencoder and classifier. The encoder of the autoencoder generates deep compressed feature maps, which contain the most important information constituting data. These maps are concatenated with feature maps generated by the classifier to classify breathing patterns. The TCNN, single-stream CNN (SCNN), and state-of-the-art classification models were applied to classify four breathing patterns: normal, slow, rapid, and breath holding. The input data consisted of chest tissue hemodynamic responses measured using a wearable near-infrared spectroscopy device on 14 healthy adult participants. Among the classification models evaluated, random forest had the lowest classification accuracy at 88.49%, while the TCNN achieved the highest classification accuracy at 94.63%. In addition, the proposed TCNN performed 2.6% better in terms of classification accuracy than an SCNN (without an autoencoder). Moreover, the TCNN mitigates the issue of declining learning performance with increasing network depth, as observed in the SCNN model. These results prove the robustness of the TCNN in classifying breathing patterns despite using a significantly smaller number of parameters and computations compared to state-of-the-art classification models.

This scoping review aimed to identify and synthesize the literature related to patient-generated health data (PGHD) among older adults with cancer in home setting. Of the 1,090 articles extracted through six databases searches, 53 were selected. Studies were published from 2007 to 2022 and the types of devices to generate PGHD included research-grade and consumer-grade wearable devices. PGHD was assessed for physical activity, vital signs, and sleep. PGHD utilization was categorized: 1) identification, monitoring, review, and analysis (100%); 2) feedback and information report (32.1%); 3) motivation (26.4%); and 4) education and coaching (17.0%). Our study reveals that various PGHDs from older adults with cancer are mainly collected passively, with limited use for interaction with healthcare providers. These results may provide valuable insights for healthcare providers into potential PGHD applications in geriatric cancer care.

OBJECTIVE: To investigate the efficacy of closed-loop acoustic stimulation (CLAS) during slow-wave sleep (SWS) to enhance slow-wave activity (SWA) and SWS in patients with Alzheimer\'s disease (AD) across multiple nights and to explore associations between stimulation, participant characteristics, and individuals\' SWS response.
METHODS: A 2-week, open-label at-home intervention study utilizing the DREEM2 headband to record sleep data and administer CLAS during SWS.
METHODS: Fifteen older patients with AD (6 women, mean age: 76.27 [SD = 6.06], mean MOCA-score: 16.07 [SD = 6.94]), living at home with their partner, completed the trial.
METHODS: Patients first wore the device for two baseline nights, followed by 14 nights during which the device was programmed to randomly either deliver acoustic stimulations of 50 ms pink noise (± 40 dB) targeted to the slow-wave up-phase during SWS or only mark the wave (sham).
RESULTS: On a group level, stimulation significantly enhanced SWA and SWS with consistent SWS enhancement throughout the intervention. However, substantial variability existed in individual responses to stimulation. Individuals received more stimulations on nights with increased SWS compared to baseline than on nights with no change or a decrease. In individuals, having lower baseline SWS correlated with receiving fewer stimulations on average during the intervention.
CONCLUSIONS: CLAS during SWS is a promising nonpharmacological method to enhance SWA and SWS in AD. However, patients with lower baseline SWS received fewer stimulations during the intervention, possibly resulting in less SWS enhancement. Individual variability in response to stimulation underscores the need to address personalized stimulation parameters in future research and therapy development.

BACKGROUND: Step counting is comparable among many research-grade and consumer-grade accelerometers in laboratory settings.
OBJECTIVE: The purpose of this study was to compare the agreement between Actical and Apple Watch step-counting in a community setting.
METHODS: Among Third Generation Framingham Heart Study participants (N=3486), we examined the agreement of step-counting between those who wore a consumer-grade accelerometer (Apple Watch Series 0) and a research-grade accelerometer (Actical) on the same days. Secondarily, we examined the agreement during each hour when both devices were worn to account for differences in wear time between devices.
RESULTS: We studied 523 participants (n=3223 person-days, mean age 51.7, SD 8.9 years; women: n=298, 57.0%). Between devices, we observed modest correlation (intraclass correlation [ICC] 0.56, 95% CI 0.54-0.59), poor continuous agreement (29.7%, n=957 of days having steps counts with ≤15% difference), a mean difference of 499 steps per day higher count by Actical, and wide limits of agreement, roughly ±9000 steps per day. However, devices showed stronger agreement in identifying who meets various steps per day thresholds (eg, at 8000 steps per day, kappa coefficient=0.49), for which devices were concordant for 74.8% (n=391) of participants. In secondary analyses, in the hours during which both devices were worn (n=456 participants, n=18,760 person-hours), the correlation was much stronger (ICC 0.86, 95% CI 0.85-0.86), but continuous agreement remained poor (27.3%, n=5115 of hours having step counts with ≤15% difference) between devices and was slightly worse for those with mobility limitations or obesity.
CONCLUSIONS: Our investigation suggests poor overall agreement between steps counted by the Actical device and those counted by the Apple Watch device, with stronger agreement in discriminating who meets certain step thresholds. The impact of these challenges may be minimized if accelerometers are used by individuals to determine whether they are meeting physical activity guidelines or tracking step counts. It is also possible that some of the limitations of these older accelerometers may be improved in newer devices.

UNASSIGNED: To assess the utility of wearable cameras in medical examinations, we created a physician-view video-based examination question and explanation, and the survey results indicated that these cameras can enhance the evaluation and educational capabilities of medical examinations.

This paper outlines the development of the \'Cardiac Abnormality Monitoring\' wearable medical device, aimed at creating a compact safety monitor integrating advanced Artificial Neural Network (ANN) algorithms. Given power consumption constraints and cost-effectiveness, a strategy combining sophisticated instruments with neural network algorithms is proposed to enhance performance. This approach aims to compete with high-end wearable devices, utilizing innovative manufacturing techniques. The paper evaluates the feasibility of employing the Levenberg-Marquardt (LM) ANN algorithm in power-conscious wearable devices, considering its potential for offline embedded systems or IoT gadgets capable of cloud-based data uploading. The Levenberg-Marquardt ANN is chosen primarily for its practicality in prototype development, with other neural network algorithms also explored to identify potential alternatives. We have compared the six neural network models and determined the model that has the potential to replace the primary neural network model. We found that the \'Kernelized SVC with PCA\' can test accuracy. To be specific, in this paper, we will evaluate the performance of the ANN model and also check its feasibility and practicality by integrating it with a constructed prototypical working model.