Jooyong Kim’s research while affiliated with Soongsil University and other places

Measuring respiratory parameters is crucial for clinical decision making and detecting abnormal patterns for disease prevention. While deep learning methods are commonly used in respiratory analysis, the image-based classification of abnormal breathing remains limited. This study developed a stitched sensor using silver-coated thread, optimized for the knit fabric’s course direction in a belt configuration. By applying a Continuous Wavelet Transform (CWT) and a two-dimension Convolutional Neural Network (2D-CNN), the model achieved 96% accuracy, with potential for further improvement through data expansion.

This study proposes a convolutional neural network (CNN)-based image analysis method to evaluate the electrical properties and uniformity of conductive fabrics treated with single-walled carbon nanotube (SWCNT) dip-coating. The conductive fabric was produced by dip-coating cotton-blended spandex with SWCNT, and the surface images were scanned and preprocessed to obtain image data, while resistance measurements were conducted to obtain labels and build the dataset. SEM analysis revealed that as the number of dip-coating cycles increased, particle density and path formation improved. The CNN model learned the relationship between surface images and resistance values, achieving a high predictive performance, with an R-squared (R²) value of 0.9422. The model demonstrated prediction accuracies of 99.1792% for the coefficient of variation (CV) of uniformly coated fabrics and 96.8877% for non-uniformly coated fabrics. Additionally, p-value analysis of all fabric samples yielded a result of 0.96044, indicating no statistically significant difference between the predicted and actual values. The proposed CNN-based model can accurately evaluate the electrical uniformity of conductive fabrics, showing potential for contributing to quality control and process optimization in production.

This study contributes to respiratory pattern detection by introducing a fabric sensor utilizing capacitance measurement and a semi-supervised machine learning algorithm known as an AI-based autoencoder. The sensor, consisting of two embroidered electrodes composed of silver-coated conductive nylon filaments, leverages the body as a dielectric material. In the research, a garment-type respiratory sensor was employed to continuously monitor respiratory data during both static (standing) and dynamic (walking, brisk walking, running) actions. The sparse autoencoder algorithm was particularly employed for individual static and dynamic actions, effectively distinguishing respiratory patterns corresponding to various movements. In addition, the sparse autoencoder helps prevent overfitting, fundamentally minimizing errors between the compression and reconstruction of signals. The maximum number of epochs was set to 2000, and the target error was set at 0.005. All data were compared against the static walking as the training baseline. Ultimately, the root mean squared error (RMSE) between static postures averaged 0.1, while the RMSEs between dynamic actions of walking, brisk walking, and running were 0.61, 0.91, and 2.78, respectively. These results suggest that movement detection through error detection is practically feasible and possesses discernible capabilities.

This paper presents an approach to pressure sensors with dual‐function resistors and capacitors implemented with interdigitated capacitors fabricated on a flexible substrate for detecting abnormal actions during walking. In this study, we proposed a highly sensitive, broad‐range pressure sensor achieved through a combination of porous Ecoflex, carbon nanotubes (CNTs), coating single‐wall carbon nanotubes (SWCNTs), and interdigitated electrodes. First, characterizations of the capacitor and resistor sensor applied onto cotton fabric are completed by precision LCR meter across the frequency at 50 kHz. Subsequently, the presence of volume fraction CNTs enhances the bond strength of composites, and coating SWCNT improves sensor sensitivity. The robustness of our presented sensor is validated through testing under high pressure (50 kPa) for more than 1000 cycles. Furthermore, the combination of CNTs and porous dielectric, along with SWCNT coating, achieves a broad detection range (500 kPa) with a sensitivity range from 0.018 (at 500 kPa) to 0.15 KPa−1 (at 5 kPa). Finally, our high‐performance resistive/capacitive pressure sensor applied on smart insoles represents a significant advancement in wearable technology for health monitoring and safety applications. Leveraging an advanced autoencoder, our proposed sensor accurately detects abnormal activity patterns, such as sudden stops or irregular gait, thereby alerting users to potential safety concerns. Its ability to detect abnormal activity patterns enhances user safety and well‐being, making it a tool for various healthcare and fitness applications.

This study aims to validate the optimal posture for the Dumbbell Biceps Curl (DBC) exercise using Textile stretch sensors and Surface Electromyography (sEMG), and then detect inaccurate posture using a Sparse Autoencoder. To validate the optimal DBC exercise posture, we measured the effects of wrist supination and the angle of the upper body and elbow on the biceps and forearm muscles. A wrist sleeve-shaped Textile stretch sensor detects wrist supination, and sEMG measures biceps and forearm muscle activation. The experiment results confirmed that an angle between the upper body and the elbow within 90°, coupled with wrist supination, constitutes the most efficient posture, maximizing the activation of the biceps while minimizing the synergistic effects of the forearm muscles. Subsequently, this posture was learned through a Sparse Autoencoder, and the Root Mean Square Error values of the trained model were lowest in the optimal posture (Biceps: 0.090, Forearm: 0.076). This suggests that Sparse Autoencoder could be useful in identifying inaccurate exercise postures. In summary, this study aims to develop an exercise posture feedback system through the integration of modern exercise physiology and technology, particularly the fusion of AI and sensor technology. The goal is to propose the potential to detect and correct inaccurate or unsafe exercise postures.

Respiratory measurement is a crucial indicator for assessing health status; however, current methods for measuring respiratory rate and frequency are passive and not intuitive. This study investigates peak detection algorithms using a resistive strain sensor integrated into a garment for respiratory rate monitoring. The sensor, constructed with CNT material and a flexible rubber substrate, exhibits high conformity to the body’s contours. Designed for respiration measurement, the sensor maintains a low 6% strain for optimal sensitivity, demonstrating a 4% decrease after 800 repetitions of 10% elongation. Garment design emphasizes cohesion between the sensor and fabric, achieved through a piping technique. Respiratory measurement relies on a resistive sensor principle, where abdominal volume changes induce tension, altering resistance. Three peak detection algorithms are evaluated: the window size algorithm, low-pass filter, and FIR filter. The window size algorithm shows a 93% matching rate for normal breathing but requires manual adjustments based on breathing speed. The low-pass filter reduces noise but introduces lag, challenging peak matching. The FIR filter effectively detects peaks at increased speeds, achieving a matching rate exceeding 98%. The study concludes that the choice of algorithm depends on respiratory scenarios, with the window size algorithm suitable for regular cycles, the low-pass filter for real-time monitoring, and the FIR filter for accelerated respiratory rates. The study primarily explores static situations, indicating the need for future research on dynamic respiratory movements to enhance algorithm versatility.

This study proposes the use of a CNN model to predict the resistance of conductive fabrics by utilizing the brightness information from their images, aiming to address the limitations of traditional contact-based measurement methods and explore the feasibility of non-contact resistance measurement. Conductive fabrics were produced using environmentally friendly cellulose fiber as a base material, with a dip-coating and padding process involving water-based single-walled carbon nanotube (SWCNT). After scanning the produced conductive fabrics and meticulously preprocessing the images, a dataset for CNN training was constructed, comprising label values corresponding to the sheet resistance of each image. ANOVA analysis confirmed a statistically significant relationship (p-value = 8.04145e^-18) between the brightness of conductive fabric images and their sheet resistance. By leveraging the relationship between the brightness of fabric images and sheet resistance, training of the CNN model yielded an RMSE of 0.0558 and an R-squared value of 0.9557, validating the effectiveness of the designed CNN model for image-based resistance prediction. This research is expected to contribute to the development of future real-time monitoring and control systems, providing a crucial foundation for the advancement of data-driven measurement and control systems based on computer vision and machine learning techniques. Furthermore, it is anticipated to unveil new possibilities for various applications of conductive fabrics.

Flexible wearable sensors are integral in diverse applications, particularly in healthcare and human–computer interaction systems. This paper introduces a resistive stretch sensor crafted from shape memory polymers (SMP) blended with carbon nanotubes (CNTs) and coated with silver paste. Initially, the sensor’s characteristics underwent evaluation using a Universal Testing Machine (UTM) and an LCR meter. These sensors showcased exceptional sensitivity, boasting a gauge factor of up to 20 at 5% strain, making them adept at detecting subtle movements or stimuli. Subsequently, the study conducted a comparison between SMP-CNT conductors with and without the silver coating layer. The durability of the sensors was validated through 1000 cycles of stretching at 4% ∆R/R0. Lastly, the sensors were utilized for monitoring respiration and measuring human breathing. Fourier transform and power spectrum density (PSD) analysis were employed to discern frequency components. Positioned between the chest and abdominal wall for contact-based respiration monitoring, the sensors revealed a dominant frequency of approximately 0.35 Hz. Signal filtering further enhanced their ability to capture respiration signals, establishing them as valuable tools for next-generation personalized healthcare applications.

Resistance training is becoming increasingly important and widespread. Decomposition of the muscle loads applied is important for injury prevention and determining the load on the targeted muscles. In this study, a flexible textile PET (polyethylene terephthalate)/SP(Spandex) SWCNT (Single-walled carbon nanotube) stretch sensor was fabricated and attached at four locations: the elbow, brachioradialis/flexor carpi radialis, biceps brachii, and triceps brachii. The stretch sensors attached to the elbow can measure the angle of elbow flexion without an IMU 9-axis sensor using quadratic fitting. A Multi-Layer Perceptron (MLP) was used to decompose the muscle volume expansions of the 3muscle by angle. The model provided a good fit for all three muscles, with R-squared values ranging from Test set 0.98725 to 0.99815. Through one input and three ouput fitting, the muscle volume expansion quantities during the bicep barbell curl were decomposed and compared with data. The results showed that the brachioradialis/flexor carpi radialis muscle maintained 13% of the arm muscle volume up to 60°, then increased to 44% at 100°. The biceps brachii muscle steadily increased to 70% from 0° up to 60°, and then maintained 40% at 100° due to the volume increase of other muscles. The triceps brachii muscle maintained 9% of the arm muscle volume up to 90°, then increased to 20% at 100°. This study shows that muscle volume expansion can be easily measured with a non-body contact wearable device, unlike many existing contact methods for measuring muscle activity like EMG (electro-myography), etc. This study provides a novel approach for easily measuring muscle volume expansion and decomposition in wearable devices, which can indirectly indicate injury prevention and muscle loading in target areas through balance optimization among local muscles.

This study aims to estimate the impact of sewing thread patterns on changes in the resistance of conductive yarns coated with silver paste. Firstly, the structure of the conductive yarns was examined, and various variations in the length and angle of individual sewing stitches were observed and analyzed through experiments. The results revealed that as the length of an individual stitch decreased, the width of the conductive yarn increased. Additionally, variations in the stitch angle resulted in different resistance values in the conductive yarn. These findings provide essential information for optimizing sewing patterns and designing components. Secondly, the comparison between models using multiple linear regression analysis and sewing neural networks was included to show optimized resistance prediction. The multiple linear regression analysis indicated that the stitch length and angle were significant variables affecting the resistance of the conductive thread. The artificial neural network model results can be valuable for optimizing sewing patterns and controlling resistance in various applications that utilize conductive thread. In addition, understanding the resistance variation in conductive thread according to sewing patterns and using optimized models to enhance component performance provides opportunities for innovation and progress. This research is necessary for the textile industry and materials engineering fields and holds high potential for practical applications in industrial settings.