This study aimed to identify symptom-based pattern clusters of Damjeok syndrome (DS) and verify differences in tongue features across distinct phenotypes, given the multitude of digestive disorders in DS. This clinical study was a prospective and observational study with 904 participants. In the first step, the DS patterns were expected to be determined using the primary categories of DS, and the hierarchical clustering based on participants' symptoms. Subsequently, three groups were divided using a K-means algorithm. Finally, a two-way analysis of variance (adjusted sex, height and weight) and Bonferroni methods were used to examine differences in tongue features among the three groups. The Commission Internationale de l'Éclairage (CIE) L^* color value in tongue coating (TC) differed significantly between the A and B groups (P = 0.050). Additionally, the TC CIE L^* color value in the center area significantly differed between the A and B groups (P = 0.012). Moreover, the TC CIE L^* color values in the root area significantly differed between the A and B groups (P = 0.009) and between the A and C groups (P = 0.044), whereas the TC CIE L^* color values in the side and tip areas were not significantly different (P = 0.149 and 0.360, respectively). No significant differences existed in TC percentage among all groups. These groupings and differences in tongue characteristics can help identify treatment options based on symptoms.

As the global population ages, the prevalence of age-related diseases is increasing. Preventing aging is necessary to reduce the burden of disease and protect population productivity. Biological age is an indicator of whether an individual is experiencing rapid, slow, or normal aging. Perceived age is highly correlated with biological age, which reflects health assessment, and is frequently used as an indicator of clinical aging. However, objective methods for quantifying perceived age are lacking. This study examines the correlations between perceived age and demographic data, facial wrinkles, and facial texture. This study used Korean medicine Data Center (KDC) data from 2017 to 2019. The study included 394 healthy men and 1,015 healthy women aged 40 years or older (range 40-58, 47.9±4.36). Facial wrinkles showed a correlation of 0.45, and facial texture variables GLCM_Information1_Mean and GLCM_Information2_Mean also showed correlations of 0.22 and -0.24, respectively. The results of this study will be helpful in developing healthcare services for middle-aged adults in Korea.

Driven by recent advances in deep learning and the growing need to decode EEG into text under open-vocabulary settings, we address the persistent issues of weak EEG-text alignment, noise sensitivity, and poor cross-subject generalization that limit prior EEG-to-Text systems. We propose Latent Diffusion with DeepConvNet (LDM-DConv), which embeds EEG into a probabilistic latent space and pretrains an encoder via a latent diffusion process with a UNet denoiser conditioned on text embeddings, thereby explicitly modeling uncertainty while strengthening cross-modal alignment. During decoding, the pretrained encoder and EEG-specialized DeepConvNet extract spatial-spectral features that are passed to a pretrained BART decoder to generate text. Experiments on the public Zurich Cognitive Language Processing (ZuCo) dataset-EEG recorded while participants read sentences-demonstrate that LDM-DConv achieves state-of-the-art performance and improved robustness to inter/intra-subject variability and noise. Quantitatively, the model attains a BLEU-1 of 42.40% and a ROUGE-1 F1 of 40.64%, surpassing strong baselines, with consistent gains across higher-order BLEU scores. These results indicate that combining diffusion-based uncertainty modeling with EEG-aware feature extraction materially improves open-vocabulary EEG-to-Text decoding. We conclude that LDM-DConv offers a robust foundation for this task and advocates future work on larger cross-subject corpora and evaluation metrics that capture semantic fidelity and generative diversity beyond exact n-gram overlap.

This study proposes an interpretable machine learning model to assist in the early-stage classification of Parkinson's disease, aiming to support early diagnosis. A stacking ensemble approach integrating Support Vector Machine (SVM), Random Forest, and XGBoost classifiers was employed on gait data obtained from the PhysioNet Gait in Parkinson's Disease dataset, consisting of 93 patients (mean age 66.3 ± 10.9 years, Hoehn & Yahr stage 2-3) and 73 healthy controls. Feature standardization and recursive feature elimination with cross-validation (RFECV) were applied to select informative gait features. The proposed model achieved 96.6% classification accuracy and an AUC of 0.990, demonstrating strong generalization and interpretability. These findings suggest that the proposed stacking ensemble framework could serve as a foundation for future predictive models for early Parkinson's detection.

Maintaining mobility is a crucial determinant of quality of life in older adults. However, due to the aging process, the functions of skeletal muscle and neuromuscular tissues begin to decline, leading to a loss of strength, balance, and reaction time. Based on the Korean National Anthropometric Data for males aged 70-84 years, the body weight was assigned to be 655.3 N. Linear static finite element analysis (FEA) was performed under quasi-static sagittal conditions. Posture angles were analyzed from 0° to 90° in 10° increments. The frame was modeled in structural steel, whereas the arm support and seat were modeled using polypropylene. A load-sharing model among the arm support (A), seat (S), and lower limbs (L) was applied with parameters A_min = 0.05, A_max = 0.40, L_max = 0.60, and α = 1.3. Total deformation and equivalent von Mises stress were computed at each angle. The walker's maximum total deformation occurred at θ = 0°(0.244 mm), while the peak equivalent stress was 23.9 MPa, which remained below the structural failure threshold. The arm support exhibited higher deformation and stress in the initial rising phase (maximum deformation = 0.111 mm), which decreased as weight shifted to the lower limbs. The seat showed the highest stress in the sitting position (0.620 MPa), which rapidly diminished toward full standing, indicating effective offloading. A transient increase in deformation was observed between 60° and 70°, but it did not compromise structural safety. The proposed device maintained structural integrity throughout the sit-to-stand (STS) transition. By formalizing arm-seat-lower-limb load sharing as design variables, this study establishes an engineering basis and verification framework for STS-assistive walkers. Future work will involve dynamic analyses and clinical validation to further assess usability and fall-prevention benefits.

Alzheimer's Disease (AD) is a progressive neurodegenerative disease characterized by memory impairment and cognitive decline. In this study, a 660 nm LED optical device was implanted under the scalp targeting the hippocampus through stereotactic surgery in 5XFAD dementia mice, and the therapeutic effects according to the LED output intensity were evaluated and optimized. The Novel Object Recognition (NOR) test and the Y-maze test were performed as behavioral experiments, and the data were quantitatively analyzed using motion tracking technology using reflective markers and a MATLAB-based data analysis format. As a result of the experiment, in the Y-maze test, short-term spatial memory was significantly improved in the 3-level output LED irradiation group, and in the NOR test, the novel object exploration time was the highest in the 4-level output. This study demonstrated the therapeutic efficacy of the stereotactic-based implantable LED device targeting the hippocampus, and suggested the potential for the treatment of Alzheimer's disease through the optimization of the LED output intensity.