Retention Time Prediction through Learning from a Small Training Data Set with a Pretrained Graph Neural Network.

Graph neural networks (GNNs) have shown remarkable performance in predicting the retention time (RT) for small molecules. However, the training data set for a particular target chromatographic system tends to exhibit scarcity, which poses a challenge because the experimental process for measuring RT is costly. To address this challenge, transfer learning has been used to leverage an abundant training data set from a related source task. In this study, we present an improved transfer learning method to better predict the RT of molecules for a target chromatographic system by learning from a small training data set with a pretrained GNN. We use a graph isomorphism network as the architecture of the GNN. The GNN is pretrained on the METLIN-SMRT data set and is then fine-tuned on the target training data set for a fixed number of training iterations using the limited-memory Broyden-Fletcher-Goldfarb-Shanno optimizer with a learning rate decay. We demonstrate that the proposed method achieves superior predictive performance on various chromatographic systems compared with that of the existing transfer learning methods, especially when only a small training data set is available for use. A potential avenue for future research is to leverage multiple small training data sets from different chromatographic systems to further enhance the generalization performance.

The protective effect of L-theanine addition to a cryoprotectant on the storage viability of lactic acid bacteria.

Lactic acid bacteria (LAB) that metabolize sugars to obtain energy and produce a large amount of lactate through the process are well known for their benefits. However, they can be used on a large scale only when good storage stability is guaranteed. The vitality and stability of several LAB strains were effectively protected in this investigation by L-theanine at 1% of the appropriate concentration (Lactiplantibacillus plantarum MG5023, Enterococcus faecium MG5232, Lactococcus lactis MG4668, Streptococcus thermophilus MG5140, and Bifidobacterium animalis subsp. lactis MG741). The inclusion of L-theanine as a protective agent significantly enhanced the viability of all strains throughout the freeze-drying process compared to that of the non-coated probiotics. The efficacy of L-theanine in improving bacterial stability and survivability was evaluated using accelerated stability tests, gastrointestinal (GI) tract survivability tests, and adhesion assays with intestinal epithelial cells. The cell surface was covered with substances including L-theanine, according to morphological findings, providing efficient defense against a variety of external stresses. Therefore, by exerting anti-freezing and anti-thawing properties, the adoption of L-theanine as a new and efficient protective agent may improve the stability and viability of a variety of probiotics.

Rainwater-Removal Image Conversion Learning with Training Pair Augmentation.

In this study, we proposed an image conversion method that efficiently removes raindrops on a camera lens from an image using a deep learning technique. The proposed method effectively presents a raindrop-removed image using the Pix2pix generative adversarial network (GAN) model, which can understand the characteristics of two images in terms of newly formed images of different domains. The learning method based on the captured image has the disadvantage that a large amount of data is required for learning and that unnecessary noise is generated owing to the nature of the learning model. In particular, obtaining sufficient original and raindrops images is the most important aspect of learning. Therefore, we proposed a method that efficiently obtains learning data by generating virtual water-drop image data and effectively identifying it using a convolutional neural network (CNN).

Tone Image Classification and Weighted Learning for Visible and NIR Image Fusion.

In this paper, to improve the slow processing speed of the rule-based visible and NIR (near-infrared) image synthesis method, we present a fast image fusion method using DenseFuse, one of the CNN (convolutional neural network)-based image synthesis methods. The proposed method applies a raster scan algorithm to secure visible and NIR datasets for effective learning and presents a dataset classification method using luminance and variance. Additionally, in this paper, a method for synthesizing a feature map in a fusion layer is presented and compared with the method for synthesizing a feature map in other fusion layers. The proposed method learns the superior image quality of the rule-based image synthesis method and shows a clear synthesized image with better visibility than other existing learning-based image synthesis methods. Compared with the rule-based image synthesis method used as the target image, the proposed method has an advantage in processing speed by reducing the processing time to three times or more.

Study on the disparate transition behaviors of the electrical/physical properties in PEDOT:PSS film depending on solvent species under a follow-up solution-treatment process.

In most solution-processed organic devices, a poly(3,4-ethylenedioxythiophene) (PEDOT) polymerized with poly(4-styrenesulfonate) (PSS) film is inevitably affected by various conditions during the subsequent solution-coating processes. To investigate the effects of direct solvent exposure on the properties of PEDOT polymerized with PSS (PEDOT:PSS) films, photoemission spectroscopy-based analytical methods were used before and after solvent-coating processes. Our results clearly indicate that PEDOT: PSS films undergo a different transition mechanism depending on the solubility of the solvent in water. The water-miscible solvents induce the solvation of hydrophilic PSS chains. As a result, this process allows the solvent to diffuse into the PEDOT: PSS film, and a conformational change between PEDOT and PSS occurs. On the other hand, the water-immiscible organic solvents cause the partial adsorption of solvent molecules at the PE surface, which leads to changes in the surface properties, including work function. Based on our finding, we demonstrate that the energy-level alignments at the organic semiconductor/electrode interface for the PEDOT: PSS films can be controlled by simple solvent treatments.

AI-driven robotic chemist for autonomous synthesis of organic molecules.

The automation of organic compound synthesis is pivotal for expediting the development of such compounds. In addition, enhancing development efficiency can be achieved by incorporating autonomous functions alongside automation. To achieve this, we developed an autonomous synthesis robot that harnesses the power of artificial intelligence (AI) and robotic technology to establish optimal synthetic recipes. Given a target molecule, our AI initially plans synthetic pathways and defines reaction conditions. It then iteratively refines these plans using feedback from the experimental robot, gradually optimizing the recipe. The system performance was validated by successfully determining synthetic recipes for three organic compounds, yielding that conversion rates that outperform existing references. Notably, this autonomous system is designed around batch reactors, making it accessible and valuable to chemists in standard laboratory settings, thereby streamlining research endeavors.

Combined Deep Learning Techniques for Mandibular Fracture Diagnosis Assistance.

Mandibular fractures are the most common fractures in dentistry. Since diagnosing a mandibular fracture is difficult when only panoramic radiographic images are used, most doctors use cone beam computed tomography (CBCT) to identify the patient's fracture location. In this study, considering the diagnosis of mandibular fractures using the combined deep learning technique, YOLO and U-Net were used as auxiliary diagnostic methods to detect the location of mandibular fractures based on panoramic images without CBCT. In a previous study, mandibular fracture diagnosis was performed using YOLO learning; in the detection performance result of the YOLOv4-based mandibular fracture diagnosis module, the precision score was approximately 97%, indicating that there was almost no misdiagnosis. In particular, fractures in the symphysis, body, angle, and ramus tend to be distributed in the middle of the mandible. Owing to the irregular fracture types and overlapping location information, the recall score was approximately 79%, which increased the detection of undiagnosed fractures. In many cases, fractures that are clearly visible to the human eye cannot be grasped. To overcome these shortcomings, the number of undiagnosed fractures can be reduced using a combination of the U-Net and YOLOv4 learning modules. U-Net is advantageous for the segmentation of fractures spread over a wide area because it performs semantic segmentation. Consequently, the undiagnosed case in the middle of the mandible, where YOLO was weak, was somewhat supplemented by the U-Net module. The precision score of the combined module was 95%, similar to that of the previous method, and the recall score improved to 87%, as the number of undiagnosed cases was reduced. Through this study, the performance of a deep learning method that can be used for the diagnosis of the mandibular bone has been improved, and it is anticipated that as an auxiliary diagnostic inspection device, it will assist dentists in making diagnoses.

Automatic Detection of Mandibular Fractures in Panoramic Radiographs Using Deep Learning.

Mandibular fracture is one of the most frequent injuries in oral and maxillo-facial surgery. Radiologists diagnose mandibular fractures using panoramic radiography and cone-beam computed tomography (CBCT). Panoramic radiography is a conventional imaging modality, which is less complicated than CBCT. This paper proposes the diagnosis method of mandibular fractures in a panoramic radiograph based on a deep learning system without the intervention of radiologists. The deep learning system used has a one-stage detection called you only look once (YOLO). To improve detection accuracy, panoramic radiographs as input images are augmented using gamma modulation, multi-bounding boxes, single-scale luminance adaptation transform, and multi-scale luminance adaptation transform methods. Our results showed better detection performance than the conventional method using YOLO-based deep learning. Hence, it will be helpful for radiologists to double-check the diagnosis of mandibular fractures.