This study presents a new method for evaluating the natural frequencies of bone-conducted sounds using spectrograms obtained by short-time Fourier transform (STFT) and machine learning (ML) and examines its validity for practical applications. The natural frequencies of bone-conducted sounds is conventionally evaluated using fast Fourier transform (FFT) analysis. Spectrograms obtained by STFT are used to visualize bone-conducted sounds as a two-dimensional time-frequency representation. Another advantage of spectrograms is that they facilitate the development of a convolutional neural network based ML model for predicting natural frequency. Forty-five healthy college students (20.36 ± 1.53 years) participated in this study, and five hammer impacts were applied to the right medial malleolus to record bone-conducted sounds generated at the right medial tibial condyle. Spectrograms were obtained from the recorded waveforms using STFT and input into a ML model trained with natural frequencies evaluated by FFT as a ground truth. The natural frequencies predicted by the ML model were compared with the natural frequencies evaluated by the FFT analysis. The mean absolute error was 12.92 ± 4.05 Hz, and the mean coefficient of determination (R2) was 0.845 ± 0.163, demonstrating the high evaluation accuracy of the proposed method. Feature analysis conducted using gradient-weighted class activation mapping revealed that the developed model focused on the peak sound pressure of the spectrogram when predicting the natural frequency. Further, it was found that the predicted values were large discrepancies from those evaluated by FFT analysis when the ML model failed to capture the peak sound pressure.
This study developed a passive exoskeleton designed to alleviate the physical burden of heavy backpacks by establishing an external load path that bypasses the body. Inspired by traditional load-resting methods, the device employs rigid structural rods to transfer the weight directly from the backpack to the ground. The exoskeleton is composed of three sections: (1) Loading section that receives the backpack load, (2) Joint section that transmits the force vertically, and (3) Grounding section integrated with torsion springs to assist ankle motion. Experiments were conducted with ten healthy participants to compare "Backpack Only" and "With Device" conditions. Due to signal quality control, electromyography (EMG) analysis was performed on seven participants (n = 7), while subjective evaluations included ten participants (n = 10). The results confirmed that the load-bypass structure significantly reduced the integrated EMG of the trapezius by an average of 45.4% (p < 0.05). This objective reduction was consistent with a dramatic improvement in subjective scores for perceived shoulder strain and backpack weight (p < 0.001). Furthermore, the torsion springs successfully assisted ankle movement during the initial swing phase, leading to a 36.1% average reduction in gastrocnemius activity (p < 0.05) without disrupting natural gait patterns. However, the prototype exhibited challenges regarding gait interference; subjective walkability and stability scores significantly decreased, and unnatural muscle activity timings were observed in the tibialis anterior and rectus femoris. These issues were attributed to two mechanical limitations: the Joint section’s lack of degrees of freedom in the frontal plane, which hindered individual variations in step width, and insufficient axial compliance in the rods, which restricted the range of leg lifting during the swing phase. Future research will focus on achieving dynamic compatibility by introducing frontal plane degrees of freedom and optimizing vertical compliance to enhance the ease of leg lifting while maintaining efficient load-support performance.
The present study presents a data-driven generation of synthetic cerebral aneurysm geometries for systematic computational hemodynamic simulations. Seven patient-specific aneurysm geometries from the right internal carotid artery were reconstructed from time-of-flight magnetic resonance angiography images and standardized through orientation alignment, followed by non-rigid registration onto a common spherical point cloud as a template. Principal component analysis (PCA) was then applied to the aligned point-cloud data to quantify morphological variability and parameterize shape deformation. The first four principal components captured over 90% of the total variance; however, higher-order components were required to capture the detailed geometrical features of the original geometries. Computational fluid dynamic (CFD) simulations were performed on the PCA-based synthetic geometries under a cavity-flow-type velocity boundary condition to investigate the influence of shape variations on intra-aneurysmal flow patterns, time-averaged wall shear stress (TAWSS), and oscillatory shear index (OSI). Independent variation of first and second principal components (PCS1 and PCS2) indicated that PCS1 primarily modulates global aneurysm aspect ratios (depth-width-height), whereas PCS2 mainly controls local surface morphology, including wrinkles and bump-like features. CFD results showed that the circulation pattern was dominated by PCS1, with higher PCS1 leading to increased TAWSS and a non-monotonic change in mean OSI that reached a minimum for near-isotropic geometries, while higher PCS2 tended to locally elevate OSI in wrinkled/bump regions. Collectively, these findings indicate that PCA-based shape parameterization provides a practical approach for generating synthetic aneurysm datasets and systematically assessing how specific morphological features govern hemodynamic behavior. The proposed approach is expected to contribute to the future development of surrogate modeling and data-driven hemodynamic prediction.
