Search

Many researchers have used sound sensors to record audio data from insects, and used these data as inputs of machine learning algorithms to classify insect species. In image classification, the convolutional neural network (CNN), a well-known deep learning algorithm, achieves better performance than any other machine learning algorithm. This performance is affected by the characteristics of the convolution filter (ConvFilter) learned inside the network. Furthermore, CNN performs well in sound classification. Unlike image classification, however, there is little research on suitable ConvFilters for sound classification. Therefore, we compare the performances of three convolution filters, 1D-ConvFilter, 3×1 2D-ConvFilter, and 3×3 2D-ConvFilter, in two different network configurations, when classifying mosquitoes using audio data. In insect sound classification, most machine learning researchers use only audio data as input. However, a classification model, which combines other information such as activity circadian rhythm, should intuitively yield improved classification results. To utilize such relevant additional information, we propose a method that defines this information as a priori probabilities and combines them with CNN outputs. Of the networks, VGG13 with 3×3 2D-ConvFilter showed the best performance in classifying mosquito species, with an accuracy of 80.8%. Moreover, adding activity circadian rhythm information to the networks showed an average performance improvement of 5.5%. The VGG13 network with 1D-ConvFilter achieved the highest accuracy of 85.7% with the additional activity circadian rhythm information.

The fiber of paper mulberry, which is mostly grown in the East Asian region, varies in both length and width—even if from the same species—depending on the country, soil, and climate where it is grown. This accounts for differences in the quality of handmade paper between domestic and imported products. Once handmade paper is manufactured as sheets, its origin cannot be determined from a morphological perspective, regardless of macroscopic and microscopic observations. Therefore, this study attempted to determine the origin of Korean and imported products using mulberry bast fiber and handmade paper from a chemometrics perspective. The inorganic components of wood are absorbed from the soil and exist in the bark of trees; thus, metal and rare-earth elements in the bast fibers and handmade papers were quantitatively analyzed using inductively coupled plasma-atomic emission spectroscopy/mass spectrometry (ICP-AES/MS). Following this, a prediction model of their origin was constructed by applying multivariate statistical analysis, i.e., a partial least squares-discriminant analysis (PLS-DA) loading plot was constructed, which helped in identifying the discrimination factors resulting from their origin. According to the ICP-AES results, Pb was only detected in the Gyeongbuk Mungyeong dak sample of the Korean bast fibers. The Chinese bast fiber sample had high contents of Ca, and B and Zn were detected only in the Chinese bast fiber samples. Bast fiber samples from China, Japan, and Thailand showed relatively high contents of Al, Ca, and Na. Meanwhile, Cu and Zn were detected in the handmade paper manufactured from the imported materials. According to the ICP-MS results, among the Korean handmade papers, the products from Gyeongbuk Mungyeong, Andong, and Chungbuk Goesan exhibited a high U content and that from Gapyeong had high Sc, Rb, Sr, and Ba contents. Thus, the elemental analysis employed in this study effectively identified the origin of the papers. The origin prediction by PLS-DA for the domestic and imported products revealed an accuracy of 86.4% and 72.7% for ICP-AES and ICP-MS methods, respectively, which is deemed acceptable to assess the origin. We believe this method significantly contributes to determining the origin of paper mulberry bast fiber and handmade paper manufactured in East Asia.

Small interfering RNA (siRNA)-based therapy has emerged as a powerful therapeutic tool due to its target-specific gene silencing. However, the therapeutic efficacy has been limited by poor cellular uptake, and therefore delivery vehicles are required to improve the cellular uptake. Herein, guanidinium-incorporated polyether-based block copolyelectrolytes were developed for improved cellular uptake and consequently increased gene silencing efficiency. Guanidinium-functionalized poly(ethylene oxide-b-allyl glycidyl ether) (G-PEO-PAGE) were prepared by living anionic ring opening polymerization of allyl glycidyl ether and post-modification to introduce guanidinium groups. G-PEO-PAGE provides efficient complexation with siRNA due to highly cationic charges of the guanidinium groups at physiological pH, resulting in significantly improved cellular uptake of siRNA and consequently increased gene silencing efficiency. In addition, G-PEO-PAGE showed significantly lower cytotoxicity compared to branched polyethylenimine as a commercial counterpart. This study demonstrates that guanidinium-incorporated block copolyelectrolytes have a great potential for efficient and low toxic siRNA delivery.

The relaxation dynamics of poly(ethyl glycidyl ether-co-isopropyl glycidyl ether-b-ethylene oxide-b-ethyl glycidyl ether-co-isopropyl glycidyl ether) triblock copolymer hydrogels were investigated ...

Depending on the different types of raw materials used to produce hanji, a Korean traditional handmade paper, there can be significant differences in the durability and mechanical properties of the final product. In this study, near-infrared spectroscopy (NIR) combined with multivariate statistical methods were used to confirm the classification possibility of hanji based on the various type of raw materials. The hanji papers were prepared from paper mulberry trees, cooking agents, and mucilage. Altogether, a total of 60 hanji spectra were collected by NIR. Then, the 60 spectra were grouped into four categories: the control, paper mulberry, cooking agent, and mucilage type based on each of the types of raw materials contained in the hanji. Three different classification algorithms – partial least squares discriminant analysis (PLS-DA), support vector machines (SVM), and random forest (RF) – were used to classify the hanji types. The best hanji material classification performance was obtained when the hanji samples were classified according to paper mulberry type, wherein the prediction accuracies of PLS-DA, SVM, and RF were 100%, 100%, and 98%, respectively. These results suggested that NIR in combination with multivariate statistical methods can be used for hanji material classification.

Complex coacervate core hydrogels were formed by simply mixing two oppositely charged polyether-based ABA triblock copolyelectrolyte solutions, where A blocks functionalized with either ammonium or...

The effects of voltage on the methane production rate (k) of various types of anaerobic sludge biomasses are quantitatively demonstrated. Voltage is supplied to the anaerobic digestion (AD) sludges independently and quickly improves the methane production rates and specific methanogenic activities (SMAs), regardless of bio-electrochemical activation of the sludge. This supports previous findings that bio-electrochemical AD (BEAD) shows higher stabilization and methane production rates than those of AD. However, in reactors fed with H2/CO2, SMA and k are not significantly increased by the voltage supply, indicating that direct electron transfer for hydrogenotrophic methanogenesis is not a major methane formation pathway. The voltage supply contributed to indirect electron transfer for H2 production. Therefore, indirect electron transfer via H2 is a significant factor in hydrogenotrophic methanogenesis. These findings indicate changes in the general pathways of AD achieved using a voltage supply and provide a better understanding of previous BEAD studies.

Algal bloom is a significant issue when managing water quality in freshwater; specifically, predicting the concentration of algae is essential to maintaining the safety of the drinking water supply system. The chlorophyll-a (Chl-a) concentration is a commonly used indicator to obtain an estimation of algal concentration. In this study, an XGBoost ensemble machine learning (ML) model was developed from eighteen input variables to predict Chl-a concentration. The composition and pretreatment of input variables to the model are important factors for improving model performance. Explainable artificial intelligence (XAI) is an emerging area of ML modeling that provides a reasonable interpretation of model performance. The effect of input variable selection on model performance was estimated, where the priority of input variable selection was determined using three indices: Shapley value (SHAP), feature importance (FI), and variance inflation factor (VIF). SHAP analysis is an XAI algorithm designed to compute the relative importance of input variables with consistency, providing an interpretable analysis for model prediction. The XGB models simulated with independent variables selected using three indices were evaluated with root mean square error (RMSE), RMSE-observation standard deviation ratio, and Nash-Sutcliffe efficiency. This study shows that the model exhibited the most stable performance when the priority of input variables was determined by SHAP. This implies that on-site monitoring can be designed to collect the selected input variables from the SHAP analysis to reduce the cost of overall water quality analysis. The independent variables were further analyzed using SHAP summary plot, force plot, target plot, and partial dependency plot to provide understandable interpretation on the performance of the XGB model. While XAI is still in the early stages of development, this study successfully demonstrated a good example of XAI application to improve the interpretation of machine learning model performance in predicting water quality.

Algal blooms have been observed worldwide and have had a serious impact on industries that use water resources, which is a problem for people and the environment. For this reason, an algae warning system is used to count the number of cyanobacterial cells and the concentration of chlorophyll-a. Several studies using multispectral or hyperspectral data to estimate chlorophyll concentration have recently been carried out. In the present study, a comparative approach was applied to estimate the concentration of chlorophyll-a at Paldang Dam, South Korea using hyperspectral data. We developed a framework for estimating chlorophyll-a using dimension reduction methods, such as principal component analysis and partial least squares, and various machine learning algorithms. We analyzed hyperspectral data collected during a field survey to locate peaks in the chlorophyll-a spectrum. The framework that used support vector regression achieved the highest R2 of 0.99, a mean square error (MSE) of 1.299 μg/cm3, and showed a small discrepancy between observed and real values relative to other frameworks. These findings suggest that by combining hyperspectral data with dimension reduction and a machine learning algorithm, it is possible to provide an accurate estimation of chlorophyll-a. Using this, chlorophyll-a can be obtained in real time through hyperspectral sensor data input from drones or unmanned aerial vehicles using the learned machine learning algorithm.

Salinization of agricultural lands, particularly rice paddies, results in the drastic decline of crop yields. Soil salinization impacts the plant physiology by inducing salt stress which may leads to osmotic stress, ionic stress and water-related nutrient imbalance. These imbalances necessitate the need for plants to produce osmolytes including proline and glycine betaine. This study aimed to elucidate the dynamic changes in proline and glycine betaine accumulation modulated by the inoculation of Brevibacterium linens RS16 in salt-sensitive and moderately salt-tolerant rice plants under salt stress conditions. This study showed the interaction of four major factors including rice genotypes with differing tolerance to salt stress, length of exposure to salt stress, level of salt stress and effects of inoculation. Salt stress resulted in significant reduction in plant growth parameters with the salt-sensitive rice genotype (IR29) having a more significant growth reduction. Both the salt-sensitive and salt-tolerant rice genotypes increased in total proline and glycine betaine accumulation at 3 days and 10 days after subjecting under 50 mM and 150 mM salt stress conditions. A significant increase in proline and glycine betaine was observed in the salt-sensitive genotype after 10 days under 50 mM and 150 mM salt stress conditions. Inoculation of the rice genotypes with B. linens RS16 resulted in the improvement of plant growth parameters in both rice genotypes, and total proline and glycine betaine accumulation, especially in IR29. This study showed that proline and glycine betaine are compatible osmolytes of rice under salt stress, and that inoculation of rice genotypes with B. linens RS16 mediated salt tolerance through improvement of plant growth parameters and proline and glycine betaine accumulation in rice plants.

Aim To examine the effect of oral diabetes medication on the risk of dementia in an elderly cohort with type 2 diabetes. Methods This was a population-based cohort study using the Korean National Health Insurance claims data from 2002 to 2013. Elderly subjects (60 years of age or older) with and without type 2 diabetes were included; patients with new-onset type 2 diabetes were further divided into the oral diabetes medication group and no-medication group. Results Among 278,290 patients with type 2 diabetes, 56,587 developed dementia (20.3%) over 11 years of follow-up. Type 2 diabetes was associated with a 1.69-fold increased risk of dementia (95% CI 1.66–1.72). Among patients with newly diagnosed type 2 diabetes, the risk of dementia was lower in the oral diabetes medication group than in the no-medication group (adjusted hazard ratio [aHR], 0.79; 95% CI 0.77–0.81). Lower risk of dementia was particularly noticeable in all of the combination therapy groups and especially lower in the combination therapy group treated with dipeptidyl peptidase 4 inhibitor (aHR 0.48, 95% CI 0.45–0.51). Conclusion Overall, the use of oral diabetes medication in type 2 diabetes patients significantly decreased the risk of dementia.

We examined the packing structure of polystyrene-coated gold nanoparticles (Au@PS) as a function of grafting density. A series of Au@PS nanoparticles with grafting densities in the range of 0.51–1.94 chains nm–2 were prepared by a ligand exchange process using thiol-terminated PS and then self-assembled at a liquid–air interface. We observed a transition from disordered to body-centered cubic (bcc) to face-centered cubic (fcc) arrangements with increasing grafting density, even though the ligand length-to-core radius ratio (λ) was as high as 3.0, a condition that typically favors nonclose-packed bcc symmetry in the self-assembly of hard nanoparticles. To explain this phenomenon, we define λeff to include the concentrated polymer brush regime as part of the “hard core”, which predicts that the softness of Au@PS nanoparticles is reduced from 1.53 to 0.14 in a theta solvent as the grafting density increases from 0.51 to 1.94 chains nm–2. This new definition of λ can also predict the effective radii of nanopa...

Artificial intelligence (AI)-based diagnostic tools have been accepted in ophthalmology. The use of retinal images, such as fundus photographs, is a promising approach for the development of AI-based diagnostic platforms. Retinal pathologies usually occur in a broad spectrum of eye diseases, including neovascular or dry age-related macular degeneration, epiretinal membrane, rhegmatogenous retinal detachment, retinitis pigmentosa, macular hole, retinal vein occlusions, and diabetic retinopathy. Here, we report a fundus image-based AI model for differential diagnosis of retinal diseases. We classified retinal images with three convolutional neural network models: ResNet50, VGG19, and Inception v3. Furthermore, the performance of several dense (fully connected) layers was compared. The prediction accuracy for diagnosis of nine classes of eight retinal diseases and normal control was 87.42% in the ResNet50 model, which added a dense layer with 128 nodes. Furthermore, our AI tool augments ophthalmologist’s performance in the diagnosis of retinal disease. These results suggested that the fundus image-based AI tool is applicable for the medical diagnosis process of retinal diseases.

The tight regulation of local auxin homeostasis and signalling maxima in xylem precursor cells specifies the organising activity of the vascular cambium and consequently promotes xylem differentiation and wood formation. However, the molecular mechanisms underlying the local auxin signalling maxima in the vascular cambium are largely unknown. Here, we reveal that brassinosteroid (BR)-activated WALLS ARE THIN1 (WAT1) facilitates wood formation by enhancing local auxin signalling in the vascular cambium in Solanum lycopersicum. Growth defects and low auxin signalling readouts in the BR-deficient tomato cultivar, Micro-Tom, were associated with a novel recessive allele, Slwat1-copi, created by the insertion of a retrotransposon in the last exon of the SlWAT1 locus. Molecular and genetic studies by generating the gain-of-function and loss-of-function tomato mutants revealed that SlWAT1 is a critical regulator for fine tuning local auxin homeostasis and signalling outputs in vascular cambium to facilitate secondary growth. Finally, we discovered that BR-regulated SlBZR1/2 directly activated downstream auxin responses by SlWAT1 upregulation in xylem precursor cells to facilitate xylem differentiation and subsequent wood formation. Our data suggest that the BR-SlBZR1/2-WAT1 signalling network contributes to the high level of auxin signalling in the vascular cambium for secondary growth.

Honeybees play a crucial role in the agriculture industry because they pollinate approximately 75% of all flowering crops. However, every year, the number of honeybees continues to decrease. Consequently, numerous researchers in various fields have persistently attempted to solve this problem. Acoustic scene classification, using sounds recorded from beehives, is an approach that can be applied to detect changes inside beehives. This method can be used to determine intervals that threaten a beehive. Currently, studies on sound analysis, using deep learning algorithms integrated with various data preprocessing methods that extract features from sound signals, continue to be conducted. However, there is little insight into how deep learning algorithms recognize audio scenes, as demonstrated by studies on image recognition. Therefore, in this study, we used a mel spectrogram, mel-frequency cepstral coefficients (MFCCs), and a constant-Q transform to compare the performance of conventional machine learning models to that of convolutional neural network (CNN) models. We used the support vector machine, random forest, extreme gradient boosting, shallow CNN, and VGG-13 models. Using gradient-weighted class activation mapping (Grad-CAM), we conducted an analysis to determine how the best-performing CNN model recognized audio scenes. The results showed that the VGG-13 model, using MFCCs as input data, demonstrated the best accuracy (91.93%). Additionally, based on the precision, recall, and F1-score for each class, we established that sounds other than those from bees were effectively recognized. Further, we conducted an analysis to determine the MFCCs that are important for classification through the visualizations obtained by applying Grad-CAM to the VGG-13 model. We believe that our findings can be used to develop a monitoring system that can consistently detect abnormal conditions in beehives early by classifying the sounds inside beehives.

In the chemical mechanical polishing process of semiconductor manufacturing, the concentration of ‘large’ particles (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge $$\end{document}≥0.5 μm) in the slurry, which is considerably larger in size than the main abrasives (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\approx $$\end{document}≈ 0.1 μm), is a critical parameter that strongly influences manufacturing defects, yields, and reliabilities of large-scale-integrated circuits. Various instruments, so-called particle counters, based on light scattering, light extinction, and holography techniques have been developed to measure and monitor the large particle concentration in semiconductor fabs in real time. However, sizeable fluctuation in the measured particle concentration complicates the statistical process control in the fabs worldwide. Here, we show that an inherent fluctuation exists in the counting of large particles, which is universal, independent of instrument type, and quantitatively determined by the instrument’s operation parameters. We analytically derive a statistical theory of the fluctuation based on Poisson statistics and validate the theory through experiments and Monte-Carlo simulation. Furthermore, we provide a strategy to enhance the measurement accuracy by statistically adjusting the instrumental parameters commonly involved in the particle counters. The present results and analyses could be useful for statistical process control in semiconductor fabs to prevent large particle-induced defects such as micro-scratches and pits on wafers.

Various microorganisms are involved in nitrogen removal, and their group compositions depend closely on operating parameters. The structures and functions of nitrification microorganisms in full-scale anaerobic-anoxic–oxic (A2/O) and oxidation ditch processes were analyzed using metagenomics and canonical correspondence analysis. The community structure of ammonia-oxidizing archaea in the oxidation ditch was 3.8 (winter) – 6.3 (summer) times higher than in A2/O, and the complete ammonia oxidizer was only found in the oxidation ditch process. The canonical correspondence analysis of various environmental variables showed that Nitrosomonadales, Crenarchaeota, and Nitrospira inopinata correlate highly with nitrification, and Nitrospira was involved in NO2−-N oxidation rather than Nitrobacter. The longer solid and hydraulic retention times in the oxidation ditch were more effective in achieving a wider range of novel nitrification than A2/O. This result indicates that microbial communities of novel nitrifiers and ammonia-oxidizing archaea improved in the oxidation ditch process, significantly contributing to stable nitrogen removal.

Many studies have attempted to predict chlorophyll-a concentrations using multiple regression models and validating them with a hold-out technique. In this study commonly used machine learning models, such as Support Vector Regression, Bagging, Random Forest, Extreme Gradient Boosting (XGBoost), Recurrent Neural Network (RNN), and Long&ndash, Short-Term Memory (LSTM), are used to build a new model to predict chlorophyll-a concentrations in the Nakdong River, Korea. We employed 1&ndash, step ahead recursive prediction to reflect the characteristics of the time series data. In order to increase the prediction accuracy, the model construction was based on forward variable selection. The fitted models were validated by means of cumulative learning and rolling window learning, as opposed to the hold&ndash, out technique. The best results were obtained when the chlorophyll-a concentration was predicted by combining the RNN model with the rolling window learning method. The results suggest that the selection of explanatory variables and 1&ndash, step ahead recursive prediction in the machine learning model are important processes for improving its prediction performance.

BR signaling pathways facilitate xylem differentiation and wood formation by fine tuning SlBZR1/SlBZR2-mediated gene expression networks involved in plant secondary growth. Brassinosteroid (BR) signaling and BR crosstalk with diverse signaling cues are involved in the pleiotropic regulation of plant growth and development. Recent studies reported the critical roles of BR biosynthesis and signaling in vascular bundle development and plant secondary growth; however, the molecular bases of these roles are unclear. Here, we performed comparative physiological and anatomical analyses of shoot morphological growth in a cultivated wild-type tomato (Solanum lycopersicum cv. BGA) and a BR biosynthetic mutant [Micro Tom (MT)]. We observed that the canonical BR signaling pathway was essential for xylem differentiation and sequential wood formation by facilitating plant secondary growth. The gradual retardation of xylem development phenotypes during shoot vegetative growth in the BR-deficient MT tomato mutant recovered completely in response to exogenous BR treatment or genetic complementation of the BR biosynthetic DWARF (D) gene. By contrast, overexpression of the tomato Glycogen synthase kinase 3 (SlGSK3) or CRISPR-Cas9 (CR)-mediated knockout of the tomato Brassinosteroid-insensitive 1 (SlBRI1) impaired BR signaling and resulted in severely defective xylem differentiation and secondary growth. Genetic modulation of the transcriptional activity of the tomato Brassinazole-resistant 1/2 (SlBZR1/SlBZR2) confirmed the positive roles of BR signaling pathways for xylem differentiation and secondary growth. Our data indicate that BR signaling pathways directly promote xylem differentiation and wood formation by canonical BR-activated SlBZR1/SlBZR2.

This study aims to compare the classification performance of statistical models on highly imbalanced kidney data. The health examination cohort database provided by the National Health Insurance Service in Korea is utilized to build models with various machine learning methods. The glomerular filtration rate (GFR) is used to diagnose chronic kidney disease (CKD). It is calculated using the Modification of Diet in Renal Disease method and classified into five stages (1, 2, 3A and 3B, 4, and 5). Different CKD stages based on the estimated GFR are considered as six classes of the response variable. This study utilizes two representative generalized linear models for classification, namely, multinomial logistic regression (multinomial LR) and ordinal logistic regression (ordinal LR), as well as two machine learning models, namely, random forest (RF) and autoencoder (AE). The classification performance of the four models is compared in terms of accuracy, sensitivity, specificity, precision, and F1-Measure. To find the best model that classifies CKD stages correctly, the data are divided into a 10-fold dataset with the same rate for each CKD stage. Results indicate that RF and AE show better performance in accuracy than the multinomial and ordinal LR models when classifying the response variable. However, when a highly imbalanced dataset is modeled, the accuracy of the model performance can distort the actual performance. This occurs because accuracy is high even if a statistical model classifies a minority class into a majority class. To solve this problem in performance interpretation, we not only consider accuracy from the confusion matrix but also sensitivity, specificity, precision, and F-1 measure for each class. To present classification performance with a single value for each model, we calculate the macro-average and micro-weighted values for each model. We conclude that AE is the best model classifying CKD stages correctly for all performance indices.

The use of deep-learning-based artificial intelligence (AI) is emerging in ophthalmology, with AI-mediated differential diagnosis of neovascular age-related macular degeneration (AMD) and dry AMD a promising methodology for precise treatment strategies and prognosis. Here, we developed deep learning algorithms and predicted diseases using 399 images of fundus. Based on feature extraction and classification with fully connected layers, we applied the Visual Geometry Group with 16 layers (VGG16) model of convolutional neural networks to classify new images. Image-data augmentation in our model was performed using Keras ImageDataGenerator, and the leave-one-out procedure was used for model cross-validation. The prediction and validation results obtained using the AI AMD diagnosis model showed relevant performance and suitability as well as better diagnostic accuracy than manual review by first-year residents. These results suggest the efficacy of this tool for early differential diagnosis of AMD in situations involving shortages of ophthalmology specialists and other medical devices.

Diatoms have traditionally been used to assess water quality; however, current research suggests that physical factors, such as habitat and landscape, may be linked to the organization of diatom assemblages in streams. The aim of this study was to determine the environmental factors affecting the physiological and ecological changes of epilithic diatoms. To this end, the dominant diatom species were used, and a strategy based on epilithic diatom habitat characteristics was investigated for river restoration. The classification and regression tree (CART) and weighted averaging (WA) regression models were used to determine the habitat preference of epilithic diatoms and physicochemical environmental factors. The 35 environmental variables and six epilithic diatom species that were dominant at 58 sites along the Han River in South Korea were used. The species abundance and composition of adherent diatoms were affected by nutrient concentration and a variety of physicochemical environmental factors. These results suggest that when evaluating water quality, various factors that affect the abundance of epilithic diatoms should be considered. Research on the autecological characteristics and environmental preferences of indicator diatom species could help establish river restoration policies and quantitative evaluation criteria for biological assessments.

Elastin-like polypeptides (ELPs) are stimulusresponsive protein-based biopolymers that exhibit phase transition behavior. Some ELP block copolymers can assemble spherical nanoparticles with thermo-sensitivity. For this reason, thermo-responsive nanoparticles derived from ELP block copolymers have been utilized in the biomedical field. In this study, four different ELP block copolymers constituted three diblocks and one triblock were synthesized. Physical properties of these self-assembled micelles were then investigated according to their block lengths and distributions. In addition, drug capturing capacities of these micelles and aggregates were explored by changing solution temperature.

The prediction of anaerobic digestion (AD) performance using numerical models, which are based on mathematics and kinetics, is being challenged by poor mechanistic understanding and the non-linear relationships between performance and operational parameters. This study demonstrated that various machine learning (ML) models using the 1-step ahead with the retraining method, which utilized AD performance data from prior states, can improve the prediction accuracy of ML models. For the four types of ML models studied, the 1-step ahead with the retraining method could improve the root mean square errors by 32–49% compared to the conventional multi-step ahead method, which was particularly noteworthy during the transition period when AD reactors were faced with loading shocks and showed inhibited methane yields. Moreover, the 1-step ahead with the retraining method showed the potential of achieving accurate predictions using a single input parameter, pH, which was considerably less labor-intensive to monitor than the other parameters often required in AD models (e.g., VSS). As such, the 1-step ahead with retraining method is suitable for efficient real-time prediction of AD performance in real-world operations.