Asteatosis is characterized by excessive moisture loss from the stratum corneum, leading to skin dryness, itching, cracking, and susceptibility to bacterial infections and allergic conditions. While subjective clinical assessments remain the standard in the diagnosis of asteatosis, recent studies have aimed to objectively evaluate skin moisture through physicochemical measurements. However, the adoption of existing research instruments for assessing skin moisture in clinical practice is limited due to their high cost, bulky design, and inconsistent measurement results. This study addresses the issue of measurement variability by developing a capacitance-based skin moisture meter with an improved load stabilizing mechanism to stabilize the contact force between the sensor and the skin, which is a primary source of measurement error. The interaction at the sensor-skin boundary area is modeled using an equivalent electrical circuit, and the optimal measurement frequency is determined to enhance accuracy and reproducibility. Based on these findings, we developed a compact capacitance-based skin moisture meter capable of measuring water content as electrostatic capacitance with low variability. Performance of the instrument was evaluated using pig skin, chosen for its physiological similarity to human skin, by calculating the correlation between capacitance readings and gravimetric water content. In addition, to assess the measurement variability, water sorption and desorption tests were conducted using filter paper and water. The water content measured by the device showed a strong linear correlation with pig dermis water content (r = 0.9296) and an inverse correlation with total water loss across whole pig skin (r = −0.9066). In water sorption and desorption tests with filter paper, the mean coefficients of variation were 1.0% and 1.9%, respectively, indicating measurement stability comparable to or higher than that of existing research instruments. Although the aforementioned investigations were limited to pig skin and filter paper, these results suggest that the novel device possesses the essential performance required for accurate, low-variability measurement of human skin moisture.

Reliable assessment of lifetime smoking exposure is essential for clinical risk stratification and public health interventions. However, current self-reported methods are subject to recall bias and misclassification. Here, we developed a new multimodal deep learning model for predicting cumulative smoking exposure based on fundus images, clinical features, and manually measured anatomic indicators. In a cohort of 8,299 subjects from the Japan Ocular Imaging Registry, pre-processing was performed. For each subject, smoking history was extracted, together with 31 anatomical features of fundus photographs and 13 clinical characteristics. The proposed model consists of three branches: (1) an image branch employing EfficientNet-B3 to extract image representations, (2) a clinical branch utilizing a residual multilayer perceptron to process clinical features, and (3) an anatomical branch using a fully connected encoder to obtain anatomical features; finally the fused multi-outputs are input to the classification-assisted multiexpert regressor to estimate Brinkman Index. Our multimodal model achieved significantly better results (MAE = 158.22, R² = 0.34) compared to unimodal models. The classifier accomplished a accuracy of 61% and an F1-score of 0.50 for smoking intervals. These observations were validated by performing ablation studies, which corroborated that the modalities complement one another and that handcrafted features improve the results. This multimodal, non-invasive, objective framework should be regarded as a methodological exploration that demonstrates the technical feasibility of estimating cumulative smoking exposure from fundus images and associated clinical data. Future studies with larger datasets and objective biomarkers are needed to further validate its clinical applicability.

Background: Accurate prediction of hemodynamic outcomes after percutaneous coronary intervention (PCI) is crucial for optimal treatment planning. We previously validated a simplified computational fluid dynamics (CFD) method for estimating fractional flow reserve (FFR) using a vascular volume ratio-based boundary condition. This study aimed to evaluate the feasibility of this method to predict post-PCI FFR based solely on pre-procedural coronary computed tomography (CT) data.

Methods: This prospective study enrolled four patients undergoing PCI. Three-dimensional coronary models were reconstructed from pre-PCI CT scans. A virtual stenting procedure was performed by expanding the stenosis according to the distal reference vessel diameter, mimicking clinical practice while not referencing the actual stent dimensions. Subsequently, post-PCI FFR_CFD was calculated using our previously established CFD method.

Results: The predicted post-PCI FFR_CFD values showed good agreement with invasively measured FFR. The diameters of the virtual stents closely approximated those of the deployed stents, with a maximum difference of 0.25 mm. However, prediction errors for post-PCI FFR were larger than those observed for pre-PCI FFR in some cases.

Conclusion: This pilot study suggests that our simplified CFD method is feasible and promising for predicting post-PCI hemodynamic outcomes using only pre-procedural anatomical information.

Larger-scale validation studies are warranted to confirm its potential as a clinical tool for non-invasive PCI planning.

Apheresis therapy requires advanced operational skills. Capabilities sufficient to respond rapidly and accurately to complications such as circuit clotting, pressure abnormalities, and air contamination are crucially important for patient safety. Nevertheless, existing training systems are often designed for centralized facilities. Moreover, they lack suitability for bedside or ICU environments, where individualized training is needed most. To address these issues, we developed a compact and portable apheresis training system that can reproduce circuit abnormalities using a magnetic particle based occlusion mechanism and an air intake unit. The 400 × 200 × 350 mm system weighs 7.6 kg, incorporating a 6% v/v magnetic particle suspension (10.7 ± 1.6 µm particle size, 5 g/cm³ density) stirred at 567.9 rpm, multiple occlusion mechanisms, a simulated arm, and a control unit. For evaluation, the device was connected to clinical blood purification equipment with blood circuits for plasma exchange (PE) and granulocyte and monocyte adsorption apheresis (GMA). Variable occlusion was achieved by adjusting the distance separating the circuit and an external magnet, yielding magnetic flux densities of 20.3 mT at 10 mm to 380.8 mT at 0 mm. Clogging rates showed high linearity (R² = 0.997, n = 10). Across 10 repeated trials for each condition, complete occlusion consistently triggered pressure alarms within a few seconds in PE and GMA systems, demonstrating high reproducibility. Characteristic pressure responses such as negative sampling pressure or inlet-outlet pressures exceeding 400 mmHg were consistent with clinically observed abnormalities. The air intake mechanism introduced air into the circuit reliably, as confirmed visually and detected using device bubble sensors. All abnormal states were reversible after deactivation, confirming system stability and durability. These findings demonstrate that this system can simulate clinically relevant circuit disturbances safely under realistic operating conditions, supporting its feasibility as a bedside training tool to enhance emergency response competencies in apheresis therapy.

Dental malocclusion is a potential contributor to chronic systemic distress; however, its underlying neural mechanisms remain unclear. This study investigated acute autonomic responses during and after tooth grinding under both appropriate and inappropriate occlusal conditions. Seventeen healthy young adults underwent tooth grinding using removable metal overlays attached to either the canines or first molars in randomized order to simulate the two occlusal dental conditions. Autonomic nervous activity was assessed by heart rate variability (HRV) analysis of electrocardiogram signals and pupillary light reflex (PLR) responses. Masseter muscle activity was recorded using surface electromyography (EMG). Both occlusal dental conditions resulted in a significant increase in EMG activity compared with the baseline resting state (p < 0.01). However, no significant differences were observed between the two occlusal conditions. HRV analysis revealed that tooth grinding significantly suppressed the parasympathetic activity compared with baseline in both occlusal conditions (p < 0.01), with no significant differences between them. Notably, a prolonged pupil redilation time following tooth grinding was observed only under inappropriate occlusal conditions relative to baseline (p < 0.01). Occlusion-specific changes in PLR were evident only under blue light stimulation, suggesting the utility of blue light as a neurological probe for melanopsin-containing intrinsically photosensitive retinal ganglion cells. These findings indicate that sympathetic recovery after grinding is diminished under malocclusion compared with natural occlusion. This altered autonomic balance may explain the association between dental malocclusions and psychological stress.

In Japanese nursing homes, diapers are typically changed at scheduled times each day. However, unnecessary changes may occur when no urination has taken place or when the diaper retains sufficient absorptive capacity. These unnecessary changes increase caregiver workload and contribute to higher costs due to the disposal of unused diapers. Monitoring urination timing and urine absorption volume can help ensure that diapers are changed only when needed, thereby improving care efficiency and reducing costs. We developed a diaper sensor system capable of detecting urination timing and measuring urine absorption volume. The system consists of comb-shaped electrodes connected to a microcomputer, with the electrodes affixed to the outside of the diaper. Urine absorption volume is derived from changes in capacitance measured between the electrodes. The previous system did not consider postures when measuring urine absorption volume. This study demonstrates the necessity of posture detection for measuring urine absorption volume and proposes a new diaper sensor system that measures urine absorption volume using posture detected by an integrated accelerometer. In phantom experiments, the proposed system reduced the mean absolute percentage error (MAPE) from 30% (previous system) to 10%, and posture detection using the accelerometer achieved 100% accuracy, demonstrating the necessity of posture detection for measuring urine absorption volume. To evaluate the practicality of the proposed system and compare its accuracy in measuring urine absorption volume with that of the previous system, a healthy adult male volunteer wore the sensor-attached diaper and accelerometer. After resting for 1 h, the subject urinated in one of three postures-supine, sitting, or 90-degree lateral-each tested eight times. The proposed system reduced the MAPE from 35% (previous) to 23%, demonstrating improved measurement accuracy in actual use scenarios.

Medical device interoperability is critical for improving healthcare quality. However, widespread adoption has been hindered by the prevalence of legacy devices lacking native communication capabilities. Replacing or retrofitting these devices is often infeasible owing to high financial costs and regulatory challenges. This study aimed to solve this problem by developing a noninvasive system that enables real-time data integration from heterogeneous medical devices. We developed a system using optical character recognition (OCR) to read numerical data from the screens of multiple medical devices using a camera or capture card. For numerical recognition, we developed two AlexNet-based models trained from scratch: a general-purpose model for standard digits, and a font-specific model for superimposed digits that occur during screen refresh transitions. The training data for the 55 classes of the font-specific model include synthetically generated images of overlapping digits created using a logical OR operation. The accuracy of the system was evaluated against frame-by-frame visual confirmation. Furthermore, its ability to perform real-time analysis was validated by integrating it with the Cartor hemodynamic simulator during a surgical procedure on a canine model of induced myocardial infarction. The general-purpose model achieved an overall accuracy of 99.84%, while the font-specific model achieved perfect accuracy of 100%, successfully resolving all cases of digit overlap. In the animal experiment, the system succeeded to track changes in cardiac function in real time, calculated from six hemodynamic parameters, with an update frequency of once every 0.63 s. The system captured the dynamic response of the left ventricular ejection curve index (SL) to multiple coronary artery ligations performed to induce myocardial infarction, where the SL value initially remained stable before showing a clear decline. This study demonstrates that an OCR-based approach can serve as a practical and effective bridge technology that enables existing non-networked medical devices to connect with advanced analytical tools such as physiological simulators, facilitating real-time clinical decision support without requiring costly equipment replacement. This paper presents a feasible method for advancing data-driven medical care in diverse clinical settings.

Defecation support in elderly care presents a major burden for both caregivers and care recipients. This places additional strain related to preparation and post-processing of defecation, and stress caused by unexpected defecation. The ultimate goal of this study was to develop a system that notifies users in advance of defecation urges. Our previous studies identified changes in impedance and phase associated with defecation. In this study, we measured the abdominal impedance and phase before and after defecation to evaluate the feasibility of predicting defecation from abdominal impedance measurements. Measurements performed at several frequencies revealed increases in impedance at higher frequencies (≥ 40 kHz) and decreases in phase prior to defecation, both of which returned to baseline levels thereafter. A comparative analysis of the impedance and phase at rest and around the time of defecation confirmed that these changes were specific to the pre-defecation period. These findings suggest the potential for a non-invasive, wearable monitoring system that could alert caregivers in advance to imminent defecation, thereby reducing the burden of unexpected defecation events in elderly care.

Recent attention has focused on the post-market retraining of AI-based software as a medical device (SaMD) following the introduction of a new approval process in Japan, known as the “Improvement Design within Approval for Timely Evaluation Notice (IDATEN).” This process facilitates the post-market performance improvement of SaMD. However, repeated use of test data raises concerns about over-adaptation to the test dataset, potentially introducing performance bias. In this study, performance bias was evaluated using a deep model designed to support hotspot detection in bone scintigraphy by simulating the selection and integration of multiple post-market models with repeated use of identical test data. Performance bias was observed in both ensemble learning approaches employing bagging- and boosting-inspired sequential aggregation with pretrained models. Additionally, bias reduction was demonstrated using Thresholdout_AUC, based on differential privacy principles. These findings are expected to be useful for SaMD development, leading to continuous performance improvements.

In stroke-induced motor paralysis, impairment of finger individuation often persists over the long term. Traditional finger function assessments in clinical rehabilitation rely on ordinal scales and subjective evaluation, limiting the ability to capture detailed recovery processes. This study proposes a novel quantitative method to evaluate finger individuation by assessing how much a single finger can move individually without involuntary movements of the other fingers. Fifty healthy adults and ten stroke patients with hemiparesis were instructed to flex each finger, starting from all fingers in fully extended position. Data from healthy participants were used to define the ranges of involuntary movement in the four non-instructed fingers during each instructed finger movement. For patient evaluation, “individuation index” was used, which was defined as the maximum flexion angle achievable within the ranges obtained from healthy participants. This index was expressed as a percentage of the maximum range of motion for each finger. In a patient with minimal motor paralysis, all fingers achieved individuation index of 100%. In patients with mild and moderate paralysis, fewer fingers achieved an individuation index of 100%, reflecting the severity of motor impairment. This method accurately detected differences in finger individuation, providing a quantitative and objective measure that could serve as a useful index for assessing motor control in stroke patients.

An increasing number of individuals in aging populations are affected by urination disorders such as benign prostatic hyperplasia and overactive bladder. Bladder diaries are widely used in the diagnosis of these disorders. Patients record the time of urination and voided volume (VV) for 3 consecutive days in their bladder diaries. A urine collection container is required for VV measurement. Inappropriate disposal or cleaning of instruments after use can result in the transmission of infection. In addition, self-recording of bladder diaries often have problems such as inaccurate recording or missing entries of urinary output. To address these issues, we developed a water-soluble paper urine collection container that can be flushed down the toilet. This system was also developed to measure VV and urinary flow rate parameters. VV, maximum urinary flow rate (Qmax), and average flow rate (Qave) were estimated based on the increase in capacitance between the electrodes placed on the exterior of the container. In a basic experiment, 0.9% saline solution was used as pseudo-urine. The relationship between the volume of pseudo-urine and change in capacitance (Δx) exhibited a strong linear correlation (R² > 0.99). A correction factor (k ≈ 1.31) was applied to account for the difference in dielectric constant between urine and saline solution. The conversion from Δx to VV was determined based on the results of the basic experiment. In a urination measurement experiment, five young males (23 ± 1 years) without symptoms of urination disorders participated. Each subject urinated ten times. Simultaneous measurements were obtained using the proposed system and a medical uroflowmeter (PicoFlow2, MEDICA S.p.A., MO, Italy). The mean errors (mean ± SD) for VV, Qmax, and Qave were 14.9 ± 41.8 mL, 2.5 ± 6.0 mL/s, and 2.0 ± 2.6 mL/s, respectively. The mean absolute percent error for VV, Qmax, and Qave were 14.1%, 15.9% and 19.9%, respectively. Paired t-tests (significance level, p < 0.01) showed no significant differences between the two measurement methods for any of the parameters. Bland-Altman analysis revealed no systematic error between the proposed system and the medical uroflowmeter for any of the items. The VV, Qmax, Qave measured using the proposed system agree well with those measured using a medical uroflowmeter. This system can be expected to be a hygienic and portable uroflowmeter that can contribute to automated bladder diary recording.

The emergence of autonomous driving is becoming a reality worldwide. However, in emergency situations, human drivers must assume control of the vehicle, hence detecting driver vigilance is critical. Although many electroencephalographic studies have documented alterations in brain rhythms associated with fatigue, the results have been inconsistent. Furthermore, in high-level automated driving contexts, the driver does not operate the vehicle except in emergencies. In this study, magnetoencephalograms from eight healthy participants were recorded as they watched a 20-minute video from the driver’s perspective. The participants were instructed to press a button when a checkered flag appeared, simulating their response to an emergency. Statistical analysis demonstrated that response time to the checkered flag was prolonged during the End period (final third) of the experiment (p = 0.04). The increase in reaction time from the Beginning (first third) to the Middle (middle third) period correlated negatively with alpha-band activities in the occipital cortices during all three periods (Beginning, r = −0.72; p = 0.04; Middle, r = −0.73, p = 0.04; End r = −0.78, p = 0.02). In other words, participants with higher alpha-band activity in the Beginning period did not experience prolonged reaction time toward the Middle period. This result suggests that alpha-band activity in the occipital cortex may serve as an indicator of whether participants maintain vigilance during autonomous driving.

Standard methods for evaluating swallowing function, such as videofluoroscopic swallowing studies and fiberoptic endoscopic evaluation, are invasive and environment-dependent. In our previous study, we confirmed through simultaneous VF measurement that specific waveforms detected by an earphone-type sensor correspond to the timing of soft palate movement during swallowing. The aim of this pilot study was to assess the feasibility and potential applicability of this sensor by examining the temporal relationship between swallowing sounds and sensor waveforms. The earphone-type sensor uses an LED and a phototransistor to detect ear canal movements via reflected infrared light. Six healthy adults swallowed 3 mL of water five times while wearing the sensor in the left ear and a throat microphone on the neck. Swallowing sounds (S I, S II, S III) and waveform points (SA, SB) were analyzed for timing. The mean SA-SB interval was 0.37 ± 0.14 s. SA preceded S I in 52.2% and S II in 100% of cases. No significant difference was observed between SA and S I, whereas SB and S II showed a significant difference (p < 0.001). These findings suggest that the earphone-type sensor waveform reflects the timing of soft palate movement prior to passage of the food bolus into the hypopharynx and esophagus. Although limited by a small sample size, this pilot study provides fundamental evidence supporting the feasibility of noninvasive swallowing assessment using an earphone-type sensor.

This study examined the effects of flat-soled footwear on toe function and muscle mass in middle-aged adults. In a two-month intervention study, participants aged 50 and older were assigned to either an intervention group that wore flat insole shoes or a control group that used standard commercial footwear. All participants maintained consistent dietary habit and activity level during the study. To assess functional outcomes, we used a custom-built toe-tapping sensor embedded in insoles, together with body composition measurements. The intervention group showed significant improvements in toe-tap frequency (36.8 vs. 30.4 taps, p < 0.05) and increases in limb muscle mass, compared to the control group. Statistical significance was confirmed using Mann-Whitney U test. The findings suggest that flat-soled shoes may enhance toe activation and maintain muscle mass, which are critical factors in preventing frailty. This accessible, non-invasive intervention may support locomotor function in aging populations. Further study is warranted to evaluate long-term benefits and broader applicability.

Cognitive training can improve specific abilities, yet little is known about how trainees adapt their visual strategies and how such adaptations relate to behavior and brain function. We followed 20 university students through a fixed, 10-session virtual reality (VR) multitasking training (one session per day, 30 minutes each) that combined a continuous target-tracking task (TTT) with an event-driven color-discrimination task (CDT). Eye movements were recorded on every session and EEG was collected pre- and post-test. Clustering of daily fixation counts across three key areas of interest uncovered three distinct gaze strategies: target-focused, distributed, and color-cue-focused that utilized peripheral vision. Strategy transitions displayed a systematic shift from distributed sampling to color-cue-focused strategy as training progressed. Behaviorally, TTT performance plateaued early, whereas improvement in CDT and the corresponding reduction in multitasking cost depended on strategy adoption, with the color-cue-focused strategy yielding the fastest reactions and highest CDT levels, while preserving high TTT levels. Neurally, the color-cue-focused group showed increased frontal-parietal theta synchrony and alpha power, whereas the target-focused group exhibited enlarged parietal P3 amplitudes and midline theta enhancement. Notably, both parietal P3 and frontal theta increases strongly correlated with CDT improvement, accounting for nearly half of the variance across participants. Our results suggest that adopting a stable visual strategy, particularly a peripheral (color-cue-focused) one, promotes resource-efficient gaze control and distinct performance-relevant neural adaptation. This points to a potential method for accelerating skill acquisition by guiding trainees towards optimal strategies and reinforcing corresponding theta/alpha-P3 signatures with real-time feedback.

Magnetoencephalography with optically pumped magnetometers (OPM-MEG) has become a focus of research. Signal sources can be estimated using OPM-MEG, where the sources are approximated as current dipoles, such as those associated with sensory evoked fields (SEFs). However, previous studies have typically used around 100 OPM-MEG sensors within high-performance magnetically shielded rooms (MSR) with active shielding. In this study, we aimed to develop OPM-MEG with a reduced number of sensors in a moderate MSR. To minimize the effects of magnetic field gradients in this environment, we fixed the sensor helmet to the MSR. Either 48 or 16 OPM-MEG sensors were used, depending on the configuration. Using these setups, we measured SEFs and compared the estimated locations of signal sources with those obtained from a conventional 306-channel MEG with a superconducting quantum interference device (SQUID-MEG). Five adults (one woman, four men), aged 33.6 ± 8.02 years (mean ± SD) participated in this experiment. To co-register the sensor positions to the participant’s head coordinates, head position indicator coils were used in the 48-channel configuration, whereas reference points embedded in the sensor helmet were used in the 16-channel configuration. Signal source estimation was performed using minimum norm estimation with dynamic statistical parametric mapping on a 5-mm grid-based volume model. SEF peak latencies were determined at the peaks of the global field power around 20 ms (N20m) from the stimulus onset. The SEF signal source locations were defined as the voxel coordinates with the highest dSPM values within the somatosensory or motor cortex at the SEF peak latencies. The SEF signal source locations estimated by the 48-channel and 16-channel OPM-MEG differed from those estimated using SQUID-MEG by 10.81 ± 5.42 mm and 11.95 ± 4.66 mm (mean ± SD), respectively. Based on previous studies comparing source estimates across multiple SQUID-MEGs, this level of discrepancy is considered acceptable. These results suggest that the source locations of SEFs can be estimated reliably even with a 16-channel OPM-MEG operating in a moderate MSR.

Measuring neural activity during communication is essential for understanding social brain processes and contributes to improving the diagnosis and treatment of communication disorders. Hyperscanning (simultaneous recording of brain activity from multiple individuals) is a highly effective method for this purpose. Magnetoencephalography (MEG) hyperscanning is particularly suitable owing to its high spatiotemporal resolution. However, conventional MEG with superconducting quantum interference devices (SQUIDs) is cumbersome and requires liquid helium to maintain superconductivity. Moreover, the rising cost of liquid helium is currently a major obstacle in SQUID-based MEG operation. In contrast, novel optically pumped magnetometer (OPM)-MEG is gaining attention because it is compact, portable, and does not require liquid helium for operation. However, OPM-MEG remains costly, and its widespread adoption is expected to take time. Therefore, until OPM-MEG becomes more widely available, a hybrid hyperscanning approach combining OPM-MEG and SQUID-MEG may be a more practical solution. In the present study, we constructed a hyperscanning system using a 48-channel OPM-MEG (HEDscan, FieldLine) and a 306-channel SQUID-MEG (Vectorview, Elekta Neuromag). Twelve pairs of adults (14 women and 10 men), aged 21.5 ± 2.7 years, participated in a turn-taking verbal communication task with meaningful words and meaningless syllables. We calculated normalized amplitudes of alpha band activity during the 2-s pre-speech interval and performed a two-way mixed-design analysis of variance (ANOVA) with MEG type (OPM and SQUID) as the between-subjects factor and condition (meaningful and meaningless) as the within-subjects factor. ANOVA revealed no significant interaction and no main effect related to MEG type [F(1, 22) = 0.171, p = 0.684; F(1, 22) = 0.061, p = 0.807, respectively]. In contrast, there was a significant main effect of condition [F(1, 22) = 11.489, p = 0.003], which indicated that the normalized amplitude of alpha band activity during the meaningless condition was larger than that of the meaningful condition. This finding is consistent with previous research using a SQUID-MEG hyperscanning system, indicating successful measurement of differing brain activities across conditions regardless of MEG sensors. This study demonstrates the feasibility of hyperscanning using OPM-MEG and SQUID-MEG.

Skin barrier function has been quantitatively evaluated through trans-epidermal water loss, which has been difficult to measure in clinical settings owing to environmental factors and the measurement time. The thickness and surface water content of the stratum corneum are important indicators of skin barrier function, and current methods for measuring these two indicators are also difficult to implement in clinical settings. Therefore, we developed a model based on skin electrical impedance to estimate the thickness and water content of the stratum corneum, enabling measurement and estimation of these two indicators in a short time. In this study, we verified this model implemented in a portable skin electrical impedance measurement device for estimating the thickness and surface water content of the stratum corneum of the skin in older adults. Thirty-four older individuals were studied. The measurement electrodes were placed in contact with the forearm skin, and an alternating signal of two frequencies was applied to measure the impedance, from which the thickness and surface water content of the stratum corneum were estimated in approximately 5 s. The correlation coefficients between the estimated and measured thickness and between the estimated and measured surface water content were 0.732 and 0.604, respectively. Furthermore, the root mean square errors of the residuals for the thickness and surface water content were 1.66 µm and 3.50 points, respectively, indicating that the model accurately estimated the thickness and surface water content of the stratum corneum, even in the skin of older adults.

Goal: The variability in soft organ shapes and positions across patients poses challenges for linear models in the reconstruction of significant local variations, while nonlinear models have difficulties with interpretability. This study aims to address these issues by proposing a mesh variational autoencoder with hierarchical latent variables (HMVAE) for 3D organ shape representation. Methods: Hierarchical latent variables capture both global and local organ features. Mesh templates ensure vertex correspondence across different resolutions. Liver and stomach meshes from 86 patients were used for training, with testing conducted in 19 patients. Results: The proposed method achieved mean vertex distances of 1.5 mm for the liver and 1.4 mm for the stomach, outperforming principal component analysis in interpolation tasks. Conclusions: The proposed HMVAE enables accurate and interpretable 3D organ reconstructions with hierarchical shape control.

Electroencephalogram (EEG) signals are critical for detecting abnormal brain activity, but their high dimensionality and complexity pose significant challenges for effective analysis. In this paper, we propose CwE-T, a novel framework that combines a channel-wise convolutional neural network (CNN)-based encoder with a single-head transformer classifier for efficient EEG abnormality detection. The channel-wise encoder compresses raw EEG signals while preserving channel independence, reducing computational costs and retaining biologically meaningful features. CwE-T was evaluated using two public datasets. For the TUH Abnormal EEG Corpus, the proposed model achieved 85.0% accuracy, 76.2% sensitivity, and 91.2% specificity at per-case level, outperforming baseline models such as EEGNet, Deep4Conv, and FusionCNN. For the CHB-MIT dataset, the proposed model achieved 85.4% sensitivity and 90.0% specificity for per-signal evaluation. Furthermore, CwE-T requires only 202M FLOPs and 2.9M parameters, making it significantly more efficient than transformer-based alternatives. The framework incorporates a channelwise design that provides potential for interpretability, offering promising directions for future research in neuroscience and clinical applications. The source code is available at https://github.com/YossiZhao/CwE-T.

This study used electroencephalography (EEG) and Representational Similarity Analysis (RSA) to investigate whether the neural representation of a concept built from a descriptive sentence was similar to the representation evoked by the name of the concept alone. Fifteen students from Kyushu University (including 1 female; mean age = 23.5, standard deviation [SD] = 2.4 years) were recruited. Two experiments were conducted in Japanese language. In Experiment 1, a stable neural representation pattern (template Representational Dissimilarity Matrix [RDM]) for occupational concepts was identified by comparing the neural patterns evoked by occupation names and their synonyms. In Experiment 2, participants read a four-word sentence to identify an occupation, followed by a congruent (ANS/T) or incongruent (ANS/F) target word. Participants' self-reported timing of concept identification was also collected. A time-resolved RSA (using Spearman's correlation) was used to track template RDM reinstatement, with significance assessed by permutation test. Event-related potentials (ERPs) associated with the final target words were analyzed using a spatiotemporal cluster-based permutation test. The ERP analysis revealed a robust P300 (P3b) component for incongruent targets relative to congruent targets (significant positive cluster at Pz: 308-464 ms, P < .05), reflecting conscious target discrimination. The RSA showed no consistent reinstatement of the template pattern at the group level. However, at the individual level, the timing of neural reinstatement was associated, although non-uniformly, with the self-reported timing of subjective concept identification. These findings suggest that neural reinstatement of knowledge during contextual understanding is not uniformly linked to external stimuli, but is coupled with internal moments of concept identification, which vary across individuals owing to different cognitive strategies.

Building on our previous findings that once-weekly high-intensity exercise, whether performed as a single bout or as three-set intervals, effectively improves cardiorespiratory function, this study investigated the effects of even lower frequency―biweekly―high-intensity interval training (HIIT) on cardiorespiratory function in healthy adults, focusing on maximal oxygen uptake (V_O2max) and ventilatory threshold (VT) as primary indicators of aerobic capacity. We hypothesized that a reduced-frequency HIIT protocol would significantly improve these cardiorespiratory function indicators over a 4-month period. Eighteen healthy adult volunteers (age: 20.7 ± 0.9 years) participated in the study. The HIIT program consisted of 8 sessions over 4-months, conducted once every two weeks. Each session included three sets of cycling at 80-90% of the maximum work rate (WR_max) until exhaustion. Cardiorespiratory function was assessed through ramp exercise tests before and after the intervention. After the 4-month HIIT program, the training group showed significant improvements in several key indicators of aerobic fitness. V_O2max, the gold standard measure of cardiorespiratory fitness, increased by +12.7 ± 10.2% (p = 0.005). WR_max during the ramp exercise test increased by 10.4 ± 5.4% (p = 0.001). VT, an important submaximal indicator of endurance capacity, also improved significantly (+14.4 ± 16.1%, p = 0.021). Progressive improvement in exercise performance was observed throughout the training period, with maximal exercise duration showing significant increases: month 1 (188 ± 69 s), month 2 (207 ± 72 s), month 3 (250 ± 102 s), and month 4 (295 ± 133 s), representing a 57.2% total improvement (ANOVA p = 0.002). Individual comparisons of pre- and post-HIIT V_O2max and VT in the training group consistently showed improvements. The control group showed no significant changes in these parameters. This study provides evidence that biweekly low-frequency HIIT can induce significant improvements in both maximal and submaximal cardiorespiratory function in healthy adults. The results suggest that substantial cardiorespiratory benefits can be achieved with much less time commitment than traditionally recommended, potentially increasing exercise adherence in time-constrained populations. Further research is needed to explore the long-term effects, underlying mechanisms, and applicability to diverse populations.

This study applies the attractor reconstruction method to photoplethysmography (PPG) signals to explore their potential in preventive health monitoring. Conventional PPG analysis methods typically rely on beat-by-beat segmentation and feature point detection, both of which are highly sensitive to noise, waveform distortion, and individual variability. To address these limitations, we applied attractor reconstruction trajectories (ACTs) as an alternative analytical approach to PPG signal analysis. ACTs utilize the entire waveform to reconstruct signal dynamics, facilitating a more holistic and noninvasive assessment of vascular health. Fifty-five participants aged 20-89 years, comprising healthy individuals and patients with chronic liver or kidney disease were studied. Three-dimensional ACTs were constructed from fingertip PPG signals and their first (DPPG) and second (SDPPG) derivatives. ACT morphologies were classified using a template-matching approach, and their distributions were compared across age groups and signal types. ACT classification revealed distinct structural characteristics across signal types. For PPG ACTs, Approximate Triangle (40.5%) and Three Points (29.5%) were the most frequent patterns, while Three Points pattern was observed only in younger and healthy middle-aged groups and exhibited the lowest uncertainty rate (14.3%). DPPG ACTs showed the greatest structural diversity, with Radiating pattern predominating in younger participants and a clear transition toward dominance of Three-lobed pattern in older groups (60.7% in participants aged 80-89 years). In contrast, SDPPG ACTs were dominated by Irregular pattern (48.6%) across all age groups and showed the highest variability and uncertainty (17.6-20.7%). These results demonstrate that ACTs can capture age-dependent differences in PPG-derived signals, with DPPG ACTs exhibiting the clearest age-related structural transitions. While limitations remain regarding sample size and classification automation, the findings support the potential of ACT-based analysis as a complementary framework for noninvasive assessment of vascular aging using PPG signals.

Accurate detection of swallowing events is essential for objective assessment of swallowing function and long-term monitoring in daily living environments. In this study, we propose an automatic method for detecting swallowing segments from neck-mounted polyvinylidene fluoride (PVDF) film signals. The PVDF film simultaneously captures swallowing sounds, mechanomyography, and laryngeal elevation with high wearability, making it suitable for continuous monitoring outside clinical settings. First, multimodal signals were acquired during 17 predefined swallowing and non-swallowing tasks, and a decision tree classifier was trained using features extracted from the time and frequency domains. The trained classifier achieved an overall accuracy of approximately 48.6% for the 17-task classification. Based on the results of this multiclass classification, the tasks were further regrouped into swallowing and non-swallowing categories, resulting in a binary classification accuracy of 92.9%. The trained model was then applied to continuous signals recorded during natural eating conditions to detect swallowing segments. Evaluation against video-based annotations demonstrated a precision of 82.5%, a recall of 88.7%, and an F1-score of 85.5%. These results indicate that swallowing segments can be reasonably detected from PVDF-based multimodal signals even under natural eating conditions involving a mixture of swallowing and various non-swallowing movements. The proposed approach provides a practical foundation for unobtrusive swallowing monitoring in daily life and long-term care settings.

Individuals exhibit a physiological and psychological response known as the stress coping response when they are exposed to external stimuli. This response represents a patterned reaction to stimuli and can be identified through fluctuations in hemodynamic parameters such as heart rate. The stress coping response is widely studied in physiological psychology, including evaluations of task-related concentration, and is expected to have practical applications in daily life. However, existing methods for measuring hemodynamic parameters typically require attaching devices to the fingers, which poses challenges for routine use due to physical constraints. Therefore, we focused on facial thermal images (FTIs) and facial near-infrared (FNI) images, which can be measured remotely using infrared thermography and near-infrared cameras, respectively. Fluctuations in skin blood flow due to changes in hemodynamic parameters are reflected in FTIs and FNIs. However, it remains unclear what specific characteristics appear in these multispectral facial images during such fluctuations. The purpose of this study was to investigate the characteristics of multispectral facial images and to evaluate response delays during fluctuations in hemodynamic parameters. We conducted stress coping response elicitation experiments and performed cross-correlation analysis between the acquired multispectral facial images and hemodynamic parameters. In typical subjects under passive coping responses, hemodynamic parameters and FTI showed a 17-second delay, with a negative correlation of approximately −0.5 in the nasal region and a positive correlation of approximately 0.64 in the cheek region. However, when averaging the correlations and time delays across all subjects under active coping responses, both FTI and FNI showed correlation values ranging from −0.10 to 0.09, indicating no significant correlation. Under passive coping responses, the correlation was weak, ranging from −0.20 to 0.14. This suggests that individual differences caused the values to offset one another when averaged. Regarding the delay, differences in wavelength were confirmed. For FNI, a delay of approximately 14 to 20 seconds was observed across the entire face under passive coping responses, while a delay of approximately 17 to 23 seconds was observed across the entire face under active coping responses. For FTI, delays in specific areas such as the nose and eye sockets were relatively short, at 1.75-4.81 seconds, while the rest of the face showed delays of approximately 15.38-37.06 seconds, indicating regional differences in delay. For FNI, no significant differences were observed in specific areas; a delay of approximately 14-23 seconds was found across the entire face. Differences in correlation and delay were dependent on the response and wavelength. Individual differences were confirmed. These results indicate that future remote measurements using facial images should select data considering wavelength properties and individual variability in correlation and delay.

Early decompression for traumatic spinal cord injury (SCI) benefits neurological recovery, yet executing timely surgery demands precise, safe bone cutting adjacent to the cord. We developed and evaluated a skills training system for cervical laminoplasty that provides real-time visual feedback on drill pressing force and spinal cord contact, using an expert-derived “teacher” waveform (median rhythm ≈ 1.06 Hz) and shadowing to guide novice performance. Twenty students without surgical experience were randomized to a training group (n = 10) or control group (n = 10). Both groups performed pre- and post-training cutting trials on standardized 3D-printed cervical models. The system measured vertical pressing force (load cell, 40 Hz) and flagged spinal cord contact via an optical threshold. Over two weeks, the training group completed six shadowing sessions with metronome guidance and on-screen warnings when exceeding force/contact thresholds; the control group received only one expert video critique during the study period. Primary outcomes were cutting time, maximum pressing force, and number of spinal cord contacts. The results were analyzed using nonparametric tests. Cutting time decreased within both groups, but was not significant (control: median time 38.0 to 28.5 s, p = 0.160; training: 47.8 to 31.7 s, p = 0.064), and the between-group difference after training was also not significant (p = 0.353). Maximum pressing force showed no significant change within groups (control: 4.2 to 3.6 N, p = 0.241; training: 3.5 to 4.0 N, p = 0.554) and also no significant difference between groups after training (p = 0.519). In contrast, post-training spinal cord contacts were significantly fewer in the training group than in the control group (median 0 vs 1.5, p = 0.006), with no significant within-group changes. This laminoplasty-specific training system prioritizes the safety goal of “cutting without contacting the cord” while maintaining efficiency and output. Visual feedback based on expert force patterns reduces spinal cord contact events and provides a reproducible, quantitative framework to accelerate safe skill acquisition in decompressive cervical surgery.

Itch is an unpleasant sensation that evokes the urge to scratch and can lead to reduced concentration, sleep disturbances, and decreased quality of life. Conventional assessment methods, such as the visual analog scale (VAS) and verbal rating scale (VRS), are widely used to evaluate itch intensity; however, these methods are subjective and may be influenced by the participants' perceptions or evaluators' interpretations. This study aimed to examine the feasibility of objectively quantifying itch intensity using transcutaneous electrical stimulation. In Experiments 1 and 2, two types of itch perception thresholds, the initial itch perception threshold (IPT) and the maximum itch perception threshold (IMT), were measured and compared with each participant's current perception threshold (CPT). Both IPT and IMT showed strong correlations (r or ρ ≥ 0.81) and positive regression relationships (R² ≥ 0.68) with CPT values. In addition, subjective evaluations revealed a strong positive correlation between VAS and VRS scores (ρ ≥ 0.61). In Experiment 3, a 20-Hz square wave electrical stimulus at a 50% duty cycle was applied while incrementally increasing the stimulation intensity from 1.2 to 2.0 times each participant's IPT. VAS scores increased significantly with increasing stimulation intensity, and statistical analysis revealed a significant main effect of stimulation level (p < 0.001). In Experiment 4, itch was induced by topical application of yam (calcium oxalate) to the skin. Itch intensity was quantified using the proposed Itch Index calculated from the ratio of IPT to CPT, as well as the VAS. A significant positive correlation was observed between the Itch Index and VAS score (r = 0.77, p = 0.01), supporting the validity of the proposed evaluation method. In contrast, no clear correlation between the Itch Index and VAS score was observed in Experiment 2, which may be attributable to inter-individual differences in sensitivity to electrically induced itch. Further investigation is required to clarify this relationship. Overall, the Itch Index based on transcutaneous electrical stimulation represents a promising approach for objective quantification of itch. This method complements conventional subjective assessment tools and may contribute to improved evaluation of pruritic symptoms.

Blood oxygen saturation (SpO₂) is one of the most important vital sign parameters. The conventional measurement method is contact-based photoplethysmography (PPG), which includes finger pulse oximetry. While PPG is usually used to measure vital signs, this method can be uncomfortable for people with sensitive skin, such as infants and critically ill individuals. Recently, research has been conducted to improve noncontact SpO₂ estimation techniques using facial videos and enhance remote photoplethysmography (r-PPG) signals to extract significant information. However, r-PPG signals are often degraded by lighting, motion, and skin tone variability. This study aimed to enhance the quality of r-PPG signal from facial videos, using feature selection to reduce complexity and overfitting, and applying random forest (RF)-based methods to model the r-PPG signals. Furthermore, we developed an innovative machine learning-based method for estimating SpO₂ level using a web camera to capture r-PPG signals from defined facial videos in a controlled laboratory environment. Initially, an AI-driven framework for face mesh detection and region-of-interest (ROI) tracker algorithm were utilized to enhance r-PPG signal quality and reduce the noise caused by ambient light and subject's motions. Face detection and ROI tracker vibration was smoothed using a Kalman filter. The resulting time-series data were processed using signal filtering. Thereafter, the RF algorithm, a supervised machine learning method, was used to predict SpO₂ values from these components. For the experiment, red-green-blue (RGB) facial video data were collected from 11 subjects with different skin tones and genders to train the RF algorithm, using a smaller dataset than those utilized in other studies. The results showed mean square error of 0.95%, root mean square error of 0.98%, mean absolute error of 0.77%, and Pearson's correlation coefficient of 0.80. Despite a small training dataset, the proposed method demonstrated notable feasibility. The RGB color values from facial videos were potentially useful for accurate estimation of SpO₂ levels in a stable environment. The proposed noncontact method is a promising alternative to traditional pulse oximetry, and has potential applications in clinical settings, particularly in remote patient monitoring, critical care monitoring, early disease detection, and telemedicine.

Turning over in bed, especially turning over at night, is a vital human unconscious behavior. Clinically, this movement disperses pressure between the body and bed, thus preventing bedsores. Several devices, such as acceleration and pressure sensors, can count turning overs automatically; however, they often require installation on the patients or in the bed. The simplest and noninvasive method to count turning overs is to record and count on video images, but this method cannot protect privacy. Images obtained using thermal sensors have been used to protect privacy; however, there are no reports of counting turning overs automatically using low-resolution sensors. We developed a novel device equipped with four low-resolution thermal sensors, with each sensor recording only an 8×8-pixel thermal image. The original data can protect patient privacy because the resolution is only ~28.8×28.8 cm per body, which is the lowest resolution compared to previous reports using thermal images. Using four sensors simultaneously enables us to collect sufficient data for automatic identification. We first used the bilinear interpolation method employed in a previous report to count turning overs; however, the results were unsatisfactory because turning overs produced extremely subtle changes in the original data compared with postural changes such as falls. After several attempts, we finally developed a unique identification program that interleaved all data from four sensors and then identified turning overs using residual neural network-18. Using the new system, the accuracy, recall, and precision of counting turning overs in bed improved to approximately 90% with an acceptable computation load in an experiment conducted on volunteers. This study demonstrated the feasibility of our device to count turning overs in clinical settings by the new identification program using four 8×8-pixel thermal images per frame, which have sufficiently low resolution to protect patient privacy.

In developed countries, the aging population is increasing steadily, raising concerns related to healthcare and patient safety. One critical issue is the growing number of bed falls and injury incidents in hospitals and nursing homes, often occurring when patients attempt to get out of bed unassisted. Accurately detecting such movements and promptly notifying nursing staff are essential for preventing accidents. This work proposes a camera-based bed fall prevention system that utilizes the YOLOv11 deep learning object detection model and a single infrared camera to detect four key patient postures: supine, side-lying, getting up, and edge-sitting. Previous research has shown that factors such as blanket occlusion and presence of caregiver complicate accurate posture recognition. To address these challenges, we developed a multilabel dataset and trained a custom YOLOv11 model capable of simultaneously detecting patient posture, head position, and bed location. Training data were collected in four different scenarios, and system evaluation was conducted using three datasets: (1) simulated laboratory data from 19 participants, (2) real hospital data from five elderly participants; and (3) hospital data captured under various lighting conditions. The system successfully identifies the “getting-up” posture, tracks head movement beyond bed boundaries, issues alerts, estimates the number of visible heads, and recognizes safety modes. Experimental results demonstrate high performance, with an average accuracy of 99.4%, sensitivity of 98.8%, and specificity of 99.7%. The proposed system offers a practical and cost-effective solution for bed fall prevention, with the potential to reduce the workload of clinical staff and improve patient safety.

This study aimed to verify whether the natural frequency of the medial gastrocnemius muscle during quiet standing―proportional to the square root of muscle stiffness―estimated from the electrically induced mechanomyogram (MMG), matched that derived from the anterior-posterior torque. Six healthy young male volunteers participated in the experiment. Each participant stood quietly on a force plate while electrical stimulation was applied to the gastrocnemius muscle via disposable surface electrodes. The electrically induced MMG and anterior-posterior torque were recorded and synchronously averaged to eliminate body-sway components from the measured signals. The averaged MMG and torque waveforms were then modeled using a single second-order system and two coupled second-order systems, respectively, and the natural frequency was estimated using a nonlinear least-squares optimization method. The estimated natural frequencies were 2.07 ± 0.15 Hz for the MMG and 2.16 ± 0.14 Hz for the torque. These findings verify that the natural frequency of the medial gastrocnemius muscle derived from the electrically induced MMG closely matches that obtained from the anterior-posterior torque, demonstrating supporting the validity of both approaches for noninvasive assessment of muscle mechanical properties during quiet standing.

Metabolic dysfunction-associated steatohepatitis (MASH) is a disease characterized by accumulation of fat droplets and fibrous tissue in the liver. This disease has attracted attention due to its high incidence and risk of severe complications. High-frequency quantitative ultrasound (QUS) has demonstrated strong potential in the diagnosis of liver diseases. However, previous studies focused only on evaluating lipid droplets in fatty liver, without accounting for the interference of other components such as fibrous tissue. This study aimed to clarify the scattered signals from various tissues by analyzing six numerical computer phantoms simulating normal liver, fatty liver and hepatitis, using an amplitude envelope statistical method, the double Nakagami (DN) model. The simulation system consisted of an ultrasound platform and a linear probe with center frequency of 31.25 MHz. Eleven plane waves ranging from −15° to +15° were transmitted and received using a compound plane-wave imaging method. The results show that the DN model matches the amplitude envelope of the original echo signals and effectively distinguishes independent signal components arising from different tissues. In fatty liver phantoms, the DN model parameter αω_F showed a strong positive correlation (r = 0.8434, p = 0.0472) with fat volume. In hepatitis phantoms, αω_F increased with increase in the fibrous tissue mixture ratio in the regions of interest, while μ_L, μ_F were close to 1, reflecting the cell distribution patterns and tissue characteristics. These results indicate that echo signals exhibit different properties when the scatterer density is high or when scattering intensity is strong, supporting the feasibility of using the DN model to differentiate fat and fibrous tissues from normal liver. However, real clinical cases are more complex, as fatty and fibrous tissue intermix in the liver, requiring further validation in the future. Nevertheless, the results show the potential of this method for real-time, noninvasive quantitative characterization of tissue properties.

This study aimed to develop a high-precision and high-speed registration method to support clinicians during minimally invasive therapies such as high-intensity focused ultrasound (HIFU) therapy. In these procedures, accurate localization and treatment of pathological regions within a restricted surgical field require a high level of expertise. Therefore, real-time and highly accurate intraoperative navigation is essential. To address this need, we proposed a method of extracting blood vessel cross-sections from intraoperative ultrasound images, quantifying their planar dispersion and structural complexity, and investigating geometric parameters that contribute to improved registration accuracy. Additionally, we examined suitable point cloud generation methods for point-based registration to enhance real-time performance. The feasibility of the proposed method was further validated under simulated clinical condition such as natural respiration. Our method achieved the HIFU-required accuracy of less than 5 mm and demonstrated processing within 0.2 seconds, which is significantly faster than conventional approaches. These findings suggest the potential applicability of our method as a surgeon assistant system in ultrasound-guided therapy.

Objective: Automatic detection of P-waves in electrocardiograms (ECGs) is difficult because of their small amplitudes and interference by ventricular activity and noise. This study aimed to improve P-wave detection using the periodic nature of sinoatrial (SA) node excitation. Methods: This study developed a phase regression deep neural network (PRDNN) that estimated the continuous phase of SA node-derived P-waves by assigning 0 rad at each P-wave onset. This architecture included bidirectional long short-term memory (BLSTM) layers, feature pyramid networks (FPN), and linear layers. The model was trained using 170 ECGs from 17 patients with second-degree atrioventricular (AV) blocks annotated by experts. Results: When tested on 22 ECGs (240 P-waves) from four AV block patients, the PRDNN achieved 98.8% recall, 99.2% precision, and F-measure of 0.990. The segmentation-based deep neural network (DNN) model used for comparison achieved 88.8% recall, 100% precision, and F-measure of 0.940. The PRDNN significantly improved sensitivity with a minimal loss in precision. Conclusion: By leveraging phase information, the proposed model improves the P-wave detection performance in automatic ECG analysis, potentially enhancing diagnostic support for clinicians.

Multitasking challenges faced by nurses are compounded by a lack of clear criteria for planning the order of nursing tasks, with the responsibility left to individual nurses. Systematizing the order of nursing tasks could facilitate the development of support tools, and appropriate education could reduce the burden on nurses. However, knowledge about systematizing task implementation order is limited. As a first step toward addressing this gap, we hypothesized that task perceptions differ between novice and experienced nurses. We therefore aimed to clarify experience-based differences in task perceptions and thought processes, as well as to identify the criteria that novice nurses use when prioritizing tasks. We surveyed nurses working in general wards. The survey included questions on 115 nursing tasks, focusing on task time, difficulty level, priority, and information used to determine execution. Valid responses were obtained from 55 nurses. When comparing nursing tasks between novice and experienced nurses, several tasks demonstrated statistically significant differences in task time, difficulty, and priority. Difficulty and priority showed a positive correlation―stronger among experienced nurses―whereas no significant difference was observed in priority and task time. Compared with experienced nurses, novice nurses tended to underestimate task time, assign more polarized difficulty levels, rely primarily on time constraints when determining task priorities, and depend on a single source of information when performing tasks. These findings contribute to a deeper understanding the complexity inherent in nursing task prioritization and highlight the challenges in developing a support tool for novice nurses.

Accurate computation of the specific absorption rate (SAR) is crucial to prevent tissue overheating and ensure compliance with radiofrequency (RF) safety regulatory standards in a magnetic resonance imaging (MRI). However, different RF transmitting magnetic (B₁⁺)-field normalization methods produce varying results, therefore, understanding their impact is vital for comparing studies, optimizing MRI protocols, and adapting to new technologies. Proper normalization also helps tailor safety assessments to individual patients with anatomical variation, ultimately improving MRI safety. We analyzed SAR values using electromagnetic simulations with four human head models (two adults and two children) within a 7T MRI birdcage coil. To account for size variation in patients, two scaled versions of each model were included, resulting in 12 total head models. We compared and analyzed RF safety by applying five B₁⁺-field normalization methods: setting the input power to 1 W and using the B₁⁺-field mean values from a single voxel, a specific volume, the entire slice, and the whole head model within the B₁⁺-field. Our findings indicate that child models exhibited higher B₁⁺-field mean values and normalized SAR_1W values (_hdSAR and _psSAR_10g) compared to adult models, primarily due to their smaller size and mass, despite identical input power. While _hdSAR decreased with increasing model size, _psSAR_10g showed no clear linear trend. Additionally, comparisons between SAR_NF and SAR_1W normalization revealed significant discrepancies, emphasizing the importance of considering model size and mass in RF exposure evaluations. This study emphasizes the importance of recognizing differences in SAR values according to normalization methods to ensure accurate quantitative analysis and MRI safety.

This study aimed to develop a method for identifying tactile sensations based on the interactions between fingertips and material surfaces using fingertip acceleration and angular velocity measurement data, and to relate the obtained data to subjective sensory ratings of tactile sensations, using machine learning. A 6-axis Inertial Measurement Unit (IMU) was attached to the fingertip of each participant, and the acceleration and angular velocity were measured while the participant traced 20 different materials. After the tracing task, the participants rated the tactile sensations on a 7-point semantic differential (SD) scale for six basic sensory items that constitute the tactile sensation: roughness, warmness, softness, wetness, friction, and comfort. The measured acceleration data were transformed into spectrograms using short-time Fourier transform (STFT). Using the spectrograms as input data, convolutional neural network (CNN) model was used for machine learning to classify each of the six sensory items into seven levels. When tested on data from new participants, the CNN model achieved classification accuracies of approximately 40-60% for all six sensory evaluation items. However, further evaluation revealed that the accuracy values for “wetness” and “warmness” largely reflected the central tendency bias in the sensory rating, resulting in reduced estimation reliability. In contrast, “friction,” “roughness,” “softness,” and “comfort” showed relatively higher estimation reliability, suggesting that the IMU data more effectively captured the characteristics of these specific sensations.

Arterial stiffness is used to diagnose atherosclerosis. The most commonly used estimation methods are indirect and non-invasive. To develop a method for direct measurement of the stiffness of blood vessels, stiffness estimation was performed on four simulated blood vessels with different stiffnesses values using a balloon catheter employed in the treatment of atherosclerosis. To measure the pressure response during balloon dilation in a simulated blood vessel, a pressurization and pressure measurement device was fabricated to pressurize the balloon catheter at a constant speed while measuring the pressure. To calculate Young's moduli of the simulated vessels, we simultaneously measured the pressure responses of the simulated blood vessels and the changes in circumferential length of the vessels using a microscope. A tensile test was also used to measure the Young's moduli of the simulated vessels. The Young's modulus values obtained by the balloon catheter method and the tensile test were compared. A linear relationship was observed between the Young's moduli determined by the two tests, indicating that the balloon catheter method can be used to estimate the Young's modulus of simulated blood vessels. To develop a simple method for estimating vessel stiffness, we calculated the reaction force received from a simulated vessel using the pressure difference during dilation in three balloon catheters with different diameters. We then evaluated whether the stiffness could be estimated by comparing the measured pressures with the Young's modulus values. Both the peak pressure after pressurization and the steady-state pressure during pressurization were related to the Young's modulus values, although the relationship depended on the diameter of the balloon catheter. Therefore, we suggest that it is possible to obtain values that reflect the Young's moduli of blood vessels using the balloon catheter method. These results suggest that with appropriate optimization of the balloon catheter diameter relative to the vessel diameter, this approach has the potential to provide pressure response values reflecting the stiffness of blood vessels.

This study reports the findings on training methods utilizing the projection-based hands-free augmented reality manual (PHM) that we developed. Eighteen participants who had completed a lecture on artificial organs and consented to participate in this study were enrolled. Participants were informed beforehand that they would assemble a hemodialysis circuit using the PHM and take tests after completing the task. Participants were divided into three groups: individual group (solo learning), conventional group (conventional group learning), and collaborative group (instructed to collaboratively assemble the circuit). After learning, participants were individually tested without using the PHM. Subsequently, a questionnaire survey was conducted to compare educational effects regarding accuracy and time required. The Kruskal-Wallis test (KW) was used to determine differences among the three groups and the Steel-Dwass test (SD) was used to compare between two groups, both with a 5% significance level. A significant difference in learning time was observed among the three groups, with a very large effect size of 0.766 (η²: KW). The median correct answer rate in the immediate post-learning test was 66.5 for the collaborative learning group, 61 for the individual learning group, and 59 for the conventional learning group. One week later, the collaborative learning group scored 64.5 (a decrease of 2 points), the individual learning group scored 59 (a decrease of 2.5 points), and the traditional learning group scored 54 (a decrease of 5 points). In the immediate post-learning test, a significant difference in correct answer rate was observed between the collaborative learning group and the traditional learning group, with a very large effect size of 1.11 (d: SD). A comparison among the three groups one week later again showed a very large effect size of 0.16 (η²: KW). No significant differences were found in the time required for the immediate and one-week post-learning tests, and in the decline in correct answer rate for the one-week post-learning test. Based on the grading trends, it may be necessary to inform students during class that “a test will be administered after class.” This would allow students to deepen their understanding of the teaching material through collaboration during the learning process.

Accurate assessment of gastric motility is essential for improving the safety and efficacy of enteral nutrition therapy in elderly patients and individuals with dysphagia undergoing percutaneous endoscopic gastrostomy (PEG). However, a reliable method for long-term, low-invasive, and stable monitoring of gastric motility is yet to be established. We developed a novel measurement system designed to monitor gastric motility by detecting intragastric pressure fluctuations through a pressure sensor affixed to a gastrostomy tube. This system conceptualizes the stomach as a single balloon, enabling real-time, low-invasive recording of pressure variations associated with gastric contractions. For preliminary evaluation, we evaluated the capability of the system to capture responses comparable to electrogastrography (EGG). We performed simultaneous 30-min recordings of EGG and intragastric pressure signals before and after nutrient infusion in a cohort of 20 PEG patients. Furthermore, we analyzed frequency- and time-frequency-domain features using fast Fourier transform, wavelet transform, and coherence analysis. Before infusion, the average dominant frequencies (DFs) of both signals were comparable, indicating no systematic bias between methods. The average coherence was 0.32 ± 0.10, suggesting a low-to-moderate correlation. Following infusion, the DF remained consistent across both methods, with a coherence of 0.30 ± 0.06. These findings suggest that the intragastric pressure signals reflect gastric motility to a degree similar to that of EGG. In conclusion, the proposed system shows the potential as a practical and effective tool for visualizing gastric motility in PEG patients.

The nurse scheduling problem (NSP) is a complex operational challenge in healthcare, with direct implications for hospital costs, staff well-being, and quality of patient care. Although many methods have been proposed, generating feasible schedules remains challenging due to the diverse and complex constraints. This study addresses a real-world NSP by formulating it as a single quadratic unconstrained binary optimization (QUBO) model, incorporating more than 15 intricate constraints derived from an active hospital ward. To rigorously enforce numerical constraints―such as total working hours―we introduce a novel slack variable method using logarithmic encoding. In addition, preprocessing nurse availability and skill data reduces problem size by eliminating infeasible assignments. A key contribution of this study is the development of a hierarchical hyperparameter optimization framework using Bayesian optimization (Optuna), which addresses the manual and critical task of parameter tuning in metaheuristics. This methodology was applied to a 32-nurse ward, generating an error-free schedule that satisfied all the hard constraints. The schedule was validated by a head nurse, confirming its practical utility. This research demonstrates the feasibility and potential of applying simulated annealing to a real-world NSP. It also establishes a practical QUBO formulation and an automated tuning pipeline, paving the way for future deployment on next-generation annealing hardware.

With advances in computer vision, medical image segmentation has become a crucial tool for diagnosis and treatment of diseases. U-Net, a widely used deep learning model, excels in extracting anatomical structures from medical images; however, optimization for multiple anatomical targets within a single framework can be challenging, particularly in cadaveric imagery with heterogeneous structures. To address this, we propose a parallel U-Net that employs multiple U-Net models to simultaneously extract various anatomical structures―arteries (abdominal aorta), veins (inferior vena cava), and organs (kidneys)―from static and dynamic (video) images. Each U-Net model is trained separately on labeled anatomical data, and their outputs are merged to generate multistructure segmentations. Experiments were conducted using abdominal images obtained from 11 Japanese cadavers. In this system, the four-layer U-Net optimized the balance between extraction accuracy and time. It achieved pixel-wise accuracy values exceeding 0.95 in most cases, even on images of cadavers not used in training. Additionally, real-time segmentation was successfully performed at over 10 fps on video frames, enabling simultaneous estimation of multiple structures despite variability across cadavers. In conclusion, the proposed parallel U-Net effectively segments multiple abdominal structures in both still and video images, offering a promising approach to advance medical imaging.

Mechanical stress applied to the lower extremities during walking is a major contributor to the development and progression of knee osteoarthritis (KOA). Adequate shock absorption during gait is essential for reducing joint loading; however, excessive midfoot pronation, characterized by deformation of the medial longitudinal arch, may impair this function. Despite its biomechanical importance, quantitative assessment of midfoot pronation during walking remains challenging because conventional motion capture systems are costly, complex, and impractical for large-scale or daily-life measurements. The purpose of this study was to develop and preliminarily validate a simple, wearable system capable of efficiently quantifying midfoot pronation during walking. The proposed system integrates five thin plantar pressure sensors with an airbag-based barometric sensor positioned over the navicular tuberosity. Instead of directly measuring joint kinematics, the system quantifies pressure-based medial midfoot deformation as a relative index of midfoot pronation. System reliability was evaluated by comparing its output with navicular displacement obtained using a three-dimensional motion capture system during repeated standing-sitting tasks, focusing on similarity in temporal patterns and relative displacement trends. In addition, dynamic midfoot pronation measurements during gait were compared with static foot structural characteristics―including navicular height and rearfoot alignment―obtained using a smartphone-based three-dimensional foot scanner. The developed system demonstrated temporal patterns of navicular displacement comparable to those recorded by motion capture, with correlation coefficients ranging from 0.55 to 0.64. During walking, midfoot pronation was captured from initial contact through terminal stance, with peak values typically observed between the loading response and midstance phases. Strong associations were identified between peak midfoot pronation output and static foot structure measures, including navicular height (r = 0.88, p = 0.05) and rearfoot alignment angle (r = −0.91, p = 0.03), consistent with established biomechanical relationships. These findings indicate that the proposed wearable system can effectively capture relative midfoot motion associated with pronation during walking. Although this pilot study had a small sample size, the results support the feasibility of the system as a practical tool for field-based gait analysis, with potential applications in screening, longitudinal monitoring, and large-scale studies of midfoot biomechanics and lower-limb joint loading.

Among the vital signs used to assess the overall condition of a patient, respiratory rate often changes earlier than heart rate and blood pressure before critical events such as thromboembolism and sepsis. Continuous respiratory rate monitoring is therefore important in both acute and chronic care settings. Small adhesive stretch sensors have emerged as a promising alternative to conventional monitoring technologies, because of their light weight and capability of localized measurements. However, the optimal anatomical placement site and orientation to maximize measurement accuracy have not been systematically investigated. In this study, we systematically evaluated the optimal placement strategy for adhesive one-axis stretch sensors (C-STRETCH^® MEASURE) by comparing six different sensor placement conditions at four anatomical locations (including chest midline and bilateral chest wall) in nine healthy male participants. Respiratory rate measured by an airflow sensor on a medical respiratory diagnostic device (PulSleep LS-140) was used as reference standard. Measurements were performed during four breathing protocols: spontaneous breathing and paced breathing at 10, 20, and 30 breaths per minute guided by metronome. Using the airflow sensor as reference, the respiratory rate error obtained from the stretch sensor under each condition and the extension rate served as evaluation metrics. Primary analysis was conducted using a linear mixed-effects model adjusted for inter-subject variability, breathing rate, and body mass index. The above-navel position produced extension rates of 8.0%-16.7%, compared with 2.9%-8.4% on the chest. Model analysis confirmed that the above-navel position yielded the lowest estimated marginal mean absolute error (0.600/min) among the six conditions in healthy men. The unadjusted Bland-Altman statistics served as supportive descriptions of agreement.