The deficiency and excess of vitamin D cause various diseases, necessitating continuous management; but it is not easy to accurately measure the serum vitamin D level in the body using a non-invasive method. The aim of this study is to investigate the correlation between vitamin D levels, body information obtained by an InBody scan, and blood parameters obtained during health checkups, to determine the optimum frequency of vitamin D quantification in the skin and to propose a vitamin D measurement method based on impedance. We assessed body composition, arm impedance, and blood vitamin D concentrations to determine the correlation between each element using multiple machine learning analyses and an algorithm which predicted the concentration of vitamin D in the body using the impedance value developed. Body fat percentage obtained from the InBody device and blood parameters albumin and lactate dehydrogenase correlated with vitamin D level. An impedance measurement frequency of 21.1 Hz was reflected in the blood vitamin D concentration at optimum levels, and a confidence level of about 75% for vitamin D in the body was confirmed. These data demonstrate that the concentration of vitamin D in the body can be predicted using impedance measurement values. This method can be used for predicting and monitoring vitamin D-related diseases and may be incorporated in wearable health measurement devices.

Malignant melanoma accounts for about 1–3% of all malignancies in the West, especially in the United States. More than 9000 people die each year. In general, it is difficult to characterize a skin lesion from a photograph. In this paper, we propose a deep learning-based computer-aided diagnostic algorithm for the classification of malignant melanoma and benign skin tumors from RGB channel skin images. The proposed deep learning model constitutes a tumor lesion segmentation model and a classification model of malignant melanoma. First, U-Net was used to classify skin lesions in dermoscopy images. We implement an algorithm to classify malignant melanoma and benign tumors using skin lesion images and expert labeling results from convolutional neural networks. The U-Net model achieved a dice similarity coefficient of 81.1% compared to the expert labeling results. The classification accuracy of malignant melanoma reached 80.06%. As a result, the proposed AI algorithm is expected to be utilized as a computer-aided diagnostic algorithm to help early detection of malignant melanoma.

In this study, we propose the direct diagnosis of thyroid cancer using a small probe. The probe can easily check the abnormalities of existing thyroid tissue without relying on experts, which reduces the cost of examining thyroid tissue and enables the initial self-examination of thyroid cancer with high accuracy. A multi-layer silicon-structured probe module is used to photograph light scattered by elastic changes in thyroid tissue under pressure to obtain a tactile image of the thyroid gland. In the thyroid tissue under pressure, light scatters to the outside depending on the presence of malignant and positive properties. A simple and easy-to-use tactile-sensation imaging system is developed by documenting the characteristics of the organization of tissues by using non-invasive technology for analyzing tactile images and judging the properties of abnormal tissues.

Facial expression recognition (FER) in the wild from various viewpoints, lighting conditions, face poses, scales, and occlusions is an extremely challenging field of research. In this study, we construct a face graph by selecting action units that play an important role in changing facial expressions, and we propose an algorithm for recognizing facial expressions using a graph convolutional network (GCN). We first generated an attention map that can highlight action units to extract important facial expression features from faces in the wild. After feature extraction, a face graph is constructed by combining the attention map with face patches, and changes in expression in the wild are recognized using a GCN. Through comparative experiments conducted using both lab-controlled and wild datasets, we prove that the proposed method is the most suitable FER approach for use with image datasets captured in the wild and those under well-controlled indoor conditions.

Visual diagnosis and rejuvenation are methods currently used to diagnose and treat pressure ulcers, respectively. However, the treatment process is difficult. We developed a biophotonic sensor to diagnose pressure ulcers and, subsequently, developed a pressure ulcer care device (PUCD.) We conducted animal and clinical trials to investigate the device’s effectiveness. We confirmed the accuracy of the pressure ulcer diagnosis algorithm to be 91% and we observed an 85% reduction in immune cells when using the PUCD to treat pressure ulcer-induced mice. Additionally, we compared the treatment group to the pressure ulcer induction group to assess the PUCD’s effectiveness in identifying immune cells through its nuclear shape. These results indicate a positive effect and suggest the use of PUCD as a recovery method for pressure ulcer diagnosis and treatment.