Search

The efficient production and sustainable operation of water electrolysis hydrogen production equipment within renewable energy systems, characterized by intermittency and volatility, necessitates dynamic simulation for the development of optimized control systems. This paper proposes the application and modification of a data-driven deep-learning model to simultaneously simulate the thermal and electrochemical responses of an alkaline water electrolyzer (AWE), yielding exceptional performance and versatility. By utilizing 4840 h of comprehensive experimental data, the causal relationships among electrolyzer operating parameters are resolved, classifying them into set values and response values for the proposed dynamic model. The model architecture is based on an attention mechanism and designed with hybrid input. Notable enhancements include embedding flow-leveraged polarization function, creating set-value cross-attention, and adopting probability-sparse attention, all of which are validated through both the validation and simulation results. Extensive exploration of various structural and training parameters leads to the selection of the best-performing model. When applied to simulation across all experimental data, the median mean percentage absolute error of the optimal model in cell voltage is a mere 1.3%, while the error in temperature is only 3.5%. Moreover, under typical operating conditions, the simulation error for both voltage and temperature remain below 2%, highlighting the outstanding dynamic modeling and simulation capabilities. These results demonstrate the potential of the proposed model for future hydrogen energy applications in dynamic control and operational optimization within renewable energy system for higher performance and better conversion efficiency. • Deep-learning model for dynamic thermoelectric simulation of water electrolyzers. • Flow-leveraged polarization function and set-value cross-attention mechanism. • Excellent accuracy, thermoelectric error of 3.5% and 1.3%, respectively. • Insights for optimizing control strategies of electrolyzer for higher efficiency. [ABSTRACT FROM AUTHOR]

Customer reviews inform potential buyers' decisions, but fake reviews in e-commerce can skew perceptions as customers may feel pressured to leave positive feedback. Detecting fake reviews in e-commerce platforms is a critical challenge, impacting online shopping and deceiving customers. Effective detection strategies, employing deep learning architectures and word embeddings, are essential to combat this issue. Specifically, the study presented in this paper employed a 1-layer Simple LSTM model, a 1D Convolutional model, and a combined CNN+LSTM model. These models were trained using different pre-trained word embeddings including Word2Vec, GloVe, FastText, and with Keras embeddings, to convert the text data into vector form. The models were evaluated based on accuracy and F1-score to provide a comprehensive measure of their performance. The results indicated that the Simple LSTM model with Word2Vec embeddings achieved an accuracy of nearly 91% and an F1-score of 0.9024, outperforming all other model-embedding combinations. The 1D convolutional model performed best without any embeddings, suggesting its ability to extract meaningful features from the raw text. The transformer-based models, BERT and DistilBERT, showed progressive learning but struggled with generalization, indicating the need for strategies such as early stopping, dropout, or regularization to prevent overfitting. Notably, the DistilBERT model consistently outperformed the LSTM model, achieving optimal performance with accuracy of 96% and an F1-score of 0.9639 using a batch size of 32 and a learning rate of 4.00E-05. [ABSTRACT FROM AUTHOR]

Customer reviews inform potential buyers' decisions, but fake reviews in e-commerce can skew perceptions as customers may feel pressured to leave positive feedback. Detecting fake reviews in e-commerce platforms is a critical challenge, impacting online shopping and deceiving customers. Effective detection strategies, employing deep learning architectures and word embeddings, are essential to combat this issue. Specifically, the study presented in this paper employed a 1-layer Simple LSTM model, a 1D Convolutional model, and a combined CNN+LSTM model. These models were trained using different pre-trained word embeddings including Word2Vec, GloVe, FastText, and Keras embeddings, to convert the text data into vector form. The models were evaluated based on accuracy and F1-score to provide a comprehensive measure of their performance. The results indicated that the Simple LSTM model with Word2Vec embeddings achieved an accuracy of nearly 91% and an F1-score of 0.9024, outperforming all other model-embedding combinations. The 1D convolutional model performed best without any embeddings, suggesting its ability to extract meaningful features from the raw text. The transformer-based models, BERT and DistilBERT, showed progressive learning but struggled with generalization, indicating the need for strategies such as early stopping, dropout, or regularization to prevent overfitting. Notably, the DistilBERT model consistently outperformed the LSTM model, achieving optimal performance with an accuracy of 96% and an F1-score of 0.9639 using a batch size of 32 and a learning rate of 4.00E-05.

Traditional Chinese medicine (TCM) possesses unique advantages in the management of blood glucose and lipids. However, there is still a significant gap in the exploration of its pharmacologically active components. Integrated strategies encompassing deep-learning prediction models and active validation based on absorbable ingredients can greatly improve the identification rate and screening efficiency in TCM. In this study, the affinity prediction of 11,549 compounds from the traditional Chinese medicine system's pharmacology database (TCMSP) with dipeptidyl peptidase-IV (DPP-IV) based on a deep-learning model was firstly conducted. With the results, Gardenia jasminoides Ellis (GJE), a food medicine with homologous properties, was selected as a model drug. The absorbed components of GJE were subsequently identified through in vivo intestinal perfusion and oral administration. As a result, a total of 38 prototypical absorbed components of GJE were identified. These components were analyzed to determine their absorption patterns after intestinal, hepatic, and systemic metabolism. Virtual docking and DPP-IV enzyme activity experiments were further conducted to validate the inhibitory effects and potential binding sites of the common constituents of deep learning and sequential metabolism. The results showed a significant DPP-IV inhibitory activity (IC50 53 ± 0.63 μg/mL) of the iridoid glycosides' potent fractions, which is a novel finding. Genipin 1-gentiobioside was screened as a promising new DPP-IV inhibitor in GJE. These findings highlight the potential of this innovative approach for the rapid screening of active ingredients in TCM and provide insights into the molecular mechanisms underlying the anti-diabetic activity of GJE. [ABSTRACT FROM AUTHOR]

Breast cancer is the most prevalent neoplasia among women, with early and accurate diagnosis critical for effective treatment. In clinical practice, however, the subjective nature of histological grading of infiltrating ductal adenocarcinoma of the breast (DAC-NOS) often leads to inconsistencies among pathologists, posing a significant challenge to achieving optimal patient outcomes. Our study aimed to address this reproducibility problem by leveraging artificial intelligence (AI). We trained a deep-learning model using a convolutional neural network-based algorithm (CNN-bA) on 100 whole slide images (WSIs) of DAC-NOS from the Cancer Genome Atlas Breast Invasive Carcinoma (TCGA-BRCA) dataset. Our model demonstrated high precision, sensitivity, and F1 score across different grading components in about 17.5 h with 19,000 iterations. However, the agreement between the model's grading and that of general pathologists varied, showing the highest agreement for the mitotic count score. These findings suggest that AI has the potential to enhance the accuracy and reproducibility of breast cancer grading, warranting further refinement and validation of this approach. [ABSTRACT FROM AUTHOR]

With the evolution of the shipping market, artificial intelligence research using ship data is being actively conducted. Smart ships and reducing ship greenhouse gas emissions are among the most actively researched topics in the maritime transport industry. Owing to the massive advances in information and communications technology, the internet of things, and big data technologies, smart ships have emerged as a very promising proposition. Numerous methodologies and network architectures can smoothly collect data from ships that are currently in operation, as is currently done in research on reducing ship fuel consumption by deep learning or conventional methods. Many extensive studies of stacked autoencoders have been carried out in the past few years. However, prior studies have not addressed the development of algorithms or deep learning-based models to classify the operating states of ships. In this paper, we propose for the first time a deep learning-based stacked autoencoder model that can classify the operating state of a ship broadly into the categories of At Sea, Stand By, and In Port, using actual ship power load data. In order to maximize the model's performance, the stacked autoencoder architecture, number of hidden layers, and number of neurons contained in each layer were measured by performance metrics such as true positive rate, false positive rate, Matthews correlation coefficient, and accuracy. It was found that the model's performance was not always improved by increasing its complexity, so the feasibility of developing and utilizing an efficient model was verified by comparing it to real data. The best-performing model had a (5–128) structure with latent layer size 9. It achieved a true positive rate of 0.9035, a false positive rate of 0.0541, a Matthews correlation coefficient of 0.9054, and an accuracy of 0.9612, clearly demonstrating that deep learning can be used to analyze ship operating modes. [ABSTRACT FROM AUTHOR]

The energy market aims to manage risks associated with prices and volatility of oil asset. It is a capital-intensive market that is rippled with chaos and complex interactions among its demand-supply derivatives. Models help users forecast such interactions, to provide investors with empirical evidence of price direction. Our study sought to investigate the reasons for the unexpected plummet in price of the energy market using evolutionary modeling - which seeks to analyze input data and yield an optimal, complete solution that are tractable, robust and low-cost with tolerance of ambiguity, uncertainty and noise. We adopt the Gabillon's model to: (a) predict spots/futures prices, (b) investigate why previous predictions failed as to why price plummet, and (c) seek to critically evaluate values reached by both proposed deep learning model and the memetic algorithm by Ojugo and Allenotor (2017). [ABSTRACT FROM AUTHOR]

Current technologies as correlation analysis, regression analysis and classification model, exhibited various limitations in the evaluation of soybean possessing potentials, including single, vague evaluation and failure of quantitative prediction, and thereby hindering more efficient and profitable soymilk industry. To solve this problem, 54 soybean cultivars and their corresponding soymilks were subjected to chemical, textural, and sensory analyses to obtain the soybean physicochemical nature (PN) and the soymilk profit and quality attribute (PQA) datasets. A deep-learning based model was established to quantitatively predict PQA data using PN data. Through 45 rounds of training with the stochastic gradient descent optimization, 9 remaining pairs of PN and PQA data were used for model validation. Results suggested that the overall prediction performance of the model showed significant improvements through iterative training, and the trained model eventually reached satisfying predictions (|relative error| ≤ 20%, standard deviation of relative error ≤ 40%) on 78% key soymilk PQAs. Future model training using big data may facilitate better prediction on soymilk odor qualities. [Display omitted] • A deep-learning based model was established for the prediction of soymilk quality. • Datasets of 54 soybeans were used for model iterative training and validation. • The extreme value-first strategy showed higher efficiency in model training. • The model output accurate and precise batch predictions on 78% quality attributes. [ABSTRACT FROM AUTHOR]

Traditional Chinese medicine (TCM) possesses unique advantages in the management of blood glucose and lipids. However, there is still a significant gap in the exploration of its pharmacologically active components. Integrated strategies encompassing deep-learning prediction models and active validation based on absorbable ingredients can greatly improve the identification rate and screening efficiency in TCM. In this study, the affinity prediction of 11,549 compounds from the traditional Chinese medicine system’s pharmacology database (TCMSP) with dipeptidyl peptidase-IV (DPP-IV) based on a deep-learning model was firstly conducted. With the results, Gardenia jasminoides Ellis (GJE), a food medicine with homologous properties, was selected as a model drug. The absorbed components of GJE were subsequently identified through in vivo intestinal perfusion and oral administration. As a result, a total of 38 prototypical absorbed components of GJE were identified. These components were analyzed to determine their absorption patterns after intestinal, hepatic, and systemic metabolism. Virtual docking and DPP-IV enzyme activity experiments were further conducted to validate the inhibitory effects and potential binding sites of the common constituents of deep learning and sequential metabolism. The results showed a significant DPP-IV inhibitory activity (IC50 53 ± 0.63 μg/mL) of the iridoid glycosides’ potent fractions, which is a novel finding. Genipin 1-gentiobioside was screened as a promising new DPP-IV inhibitor in GJE. These findings highlight the potential of this innovative approach for the rapid screening of active ingredients in TCM and provide insights into the molecular mechanisms underlying the anti-diabetic activity of GJE.

Optical coherence tomography (OCT) has become the best imaging tool to assess calcified plaque and nodule. However, every OCT pullback has numerous images, and artificial analysis requires too much time and energy. Thus, it is unsuitable for clinical application. This study aimed to develop and validate an automatic assessment of calcified plaque and nodule by OCT using deep-learning model. The OCT images of calcified plaque and nodule were labeled by two expert readers based on the consensus. A deep-learning model with a MultiScale and MultiTask u-net network (MS-MT u-net) was developed. Then, with the ground truth labeled by expert readers as reference, the diagnostic accuracy and agreement of the model was validated. For the pixelwise evaluation of calcified plaque, the model had a high performance with precision (93.95%), recall (88.95%), and F1 score (91.38%). For the lesion-level evaluation of calcified plaque, the quantitative metrics by the model excellently correlated with the ground truth (calcium score, r = 0.90, p < 0.01; calcified volume, r = 0.99, p < 0.01). For calcified nodules, the model showed excellent diagnostic performance including sensitivity (91.7%), specificity (89.3%), and accuracy (91.0%). We developed a novel deep-learning model to identify the attributes of calcified plaque and nodule. This model provided excellent diagnostic accuracy and agreement with the ground truth, thereby reducing the subjectivity of manual measurements and substantially saving time. These findings can help practitioners efficiently adopt appropriate therapeutic strategies to treat calcified lesions. [ABSTRACT FROM AUTHOR]

Breast cancer is the most prevalent neoplasia among women, with early and accurate diagnosis critical for effective treatment. In clinical practice, however, the subjective nature of histological grading of infiltrating ductal adenocarcinoma of the breast (DAC-NOS) often leads to inconsistencies among pathologists, posing a significant challenge to achieving optimal patient outcomes. Our study aimed to address this reproducibility problem by leveraging artificial intelligence (AI). We trained a deep-learning model using a convolutional neural network-based algorithm (CNN-bA) on 100 whole slide images (WSIs) of DAC-NOS from the Cancer Genome Atlas Breast Invasive Carcinoma (TCGA-BRCA) dataset. Our model demonstrated high precision, sensitivity, and F1 score across different grading components in about 17.5 h with 19,000 iterations. However, the agreement between the model’s grading and that of general pathologists varied, showing the highest agreement for the mitotic count score. These findings suggest that AI has the potential to enhance the accuracy and reproducibility of breast cancer grading, warranting further refinement and validation of this approach.

There is a need to identify biomarkers predictive of response to neoadjuvant chemotherapy (NACT) in triple-negative breast cancer (TNBC). We previously obtained evidence that a polyamine signature in the blood is associated with TNBC development and progression. In this study, we evaluated whether plasma polyamines and other metabolites may identify TNBC patients who are less likely to respond to NACT. Pre-treatment plasma levels of acetylated polyamines were elevated in TNBC patients that had moderate to extensive tumor burden (RCB-II/III) following NACT compared to those that achieved a complete pathological response (pCR/RCB-0) or had minimal residual disease (RCB-I). We further applied artificial intelligence to comprehensive metabolic profiles to identify additional metabolites associated with treatment response. Using a deep learning model (DLM), a metabolite panel consisting of two polyamines as well as nine additional metabolites was developed for improved prediction of RCB-II/III. The DLM has potential clinical value for identifying TNBC patients who are unlikely to respond to NACT and who may benefit from other treatment modalities.

Self-driving vehicles require a number of tests to prevent fatal accidents and ensure their appropriate operation in the physical world. However, conducting vehicle tests on the road is difficult because such tests are expensive and labor intensive. In this study, we used an autonomous-driving simulator, and investigated the three-dimensional environmental perception problem of the simulated system. Using the open-source CARLA simulator, we generated a CarlaSim from unreal traffic scenarios, comprising 15 000 camera-LiDAR (Light Detection and Ranging) samples with annotations and calibration files. Then, we developed Multi-Sensor Fusion Perception (MSFP) model for consuming two-modal data and detecting objects in the scenes. Furthermore, we conducted experiments on the KITTI and CarlaSim datasets; the results demonstrated the effectiveness of our proposed methods in terms of perception accuracy, inference efficiency, and generalization performance. The results of this study will faciliate the future development of autonomous-driving simulated tests.

With the evolution of the shipping market, artificial intelligence research using ship data is being actively conducted. Smart ships and reducing ship greenhouse gas emissions are among the most actively researched topics in the maritime transport industry. Owing to the massive advances in information and communications technology, the internet of things, and big data technologies, smart ships have emerged as a very promising proposition. Numerous methodologies and network architectures can smoothly collect data from ships that are currently in operation, as is currently done in research on reducing ship fuel consumption by deep learning or conventional methods. Many extensive studies of stacked autoencoders have been carried out in the past few years. However, prior studies have not addressed the development of algorithms or deep learning-based models to classify the operating states of ships. In this paper, we propose for the first time a deep learning-based stacked autoencoder model that can classify the operating state of a ship broadly into the categories of At Sea, Stand By, and In Port, using actual ship power load data. In order to maximize the model’s performance, the stacked autoencoder architecture, number of hidden layers, and number of neurons contained in each layer were measured by performance metrics such as true positive rate, false positive rate, Matthews correlation coefficient, and accuracy. It was found that the model’s performance was not always improved by increasing its complexity, so the feasibility of developing and utilizing an efficient model was verified by comparing it to real data. The best-performing model had a (5–128) structure with latent layer size 9. It achieved a true positive rate of 0.9035, a false positive rate of 0.0541, a Matthews correlation coefficient of 0.9054, and an accuracy of 0.9612, clearly demonstrating that deep learning can be used to analyze ship operating modes.

With the advancement in technology, machine learning can be applied to diagnose the mass/tumor in the brain using magnetic resonance imaging (MRI). This work proposes a novel developed transfer deep-learning model for the early diagnosis of brain tumors into their subclasses, such as pituitary, meningioma, and glioma. First, various layers of isolated convolutional-neural-network (CNN) models are built from scratch to check their performances for brain MRI images. Then, the 22-layer, binary-classification (tumor or no tumor) isolated-CNN model is re-utilized to re-adjust the neurons’ weights for classifying brain MRI images into tumor subclasses using the transfer-learning concept. As a result, the developed transfer-learned model has a high accuracy of 95.75% for the MRI images of the same MRI machine. Furthermore, the developed transfer-learned model has also been tested using the brain MRI images of another machine to validate its adaptability, general capability, and reliability for real-time application in the future. The results showed that the proposed model has a high accuracy of 96.89% for an unseen brain MRI dataset. Thus, the proposed deep-learning framework can help doctors and radiologists diagnose brain tumors early.

There is a need to identify biomarkers predictive of response to neoadjuvant chemotherapy (NACT) in triple-negative breast cancer (TNBC). We previously obtained evidence that a polyamine signature in the blood is associated with TNBC development and progression. In this study, we evaluated whether plasma polyamines and other metabolites may identify TNBC patients who are less likely to respond to NACT. Pre-treatment plasma levels of acetylated polyamines were elevated in TNBC patients that had moderate to extensive tumor burden (RCB-II/III) following NACT compared to those that achieved a complete pathological response (pCR/RCB-0) or had minimal residual disease (RCB-I). We further applied artificial intelligence to comprehensive metabolic profiles to identify additional metabolites associated with treatment response. Using a deep learning model (DLM), a metabolite panel consisting of two polyamines as well as nine additional metabolites was developed for improved prediction of RCB-II/III. The DLM has potential clinical value for identifying TNBC patients who are unlikely to respond to NACT and who may benefit from other treatment modalities., Competing Interests: Author SM declares honoraria from Novartis, Pfizer and research funding from Oncothyreon (Inst), Pfizer (Inst), Novartis (Inst), Genentech (Inst), Takeda (Inst), Bayer (Inst), EMD Serono (Inst), Genentech (Inst); travel, accommodations, expenses: Novartis, Pfizer. Author JLi declares consulting or advisory role: Pfizer, AstraZeneca, Medivation/Pfizer; speakers' bureau from Physician Education Resource, UpToDate, Med Learning, Group, Medscape; research funding from Novartis (Inst), Bristol-Myers Squibb (Inst), Genentech (Inst), Pfizer (Inst), EMD Serono (Inst), Jounce Therapeutics (Inst), GlaxoSmithKline (Inst), Medivation/Pfizer (Inst); patents, royalties, other intellectual property: UpToDate; travel, accommodations, expenses: Physician Education Resource, Med Learning Group, Medscape. Author DT declares consulting or advisory role: AstraZeneca, Genomic Health, Gilead Sciences Inc, GlaxoSmithKline, Novartis Pharma, OncoPep, Pfizer; research funding: Pfizer, Novartis Pharma. Author VV declares honoraria: Genentec; consulting or advisory role: Genentech; travel, accommodations, expenses: Genentech. Author SD declares honoraria: Novartis; consulting or advisory role: Tempus, Taiho Pharmaceutical, Pfizer; research funding: EMD Serono, Guardant Health; travel, accommodations; expenses: Phillips Gilmore Oncology Communications. Author BAd declares consulting or advisory role: Bright Pink, AbbVie; research funding: AbbVie (Inst), PharmaMar (Inst), AstraZeneca (Inst), InVitae (Inst); travel, accommodations, expenses: AstraZeneca. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2022 Irajizad, Wu, Vykoukal, Murage, Spencer, Dennison, Moulder, Ravenberg, Lim, Litton, Tripathym, Valero, Damodaran, Rauch, Adrada, Candelaria, White, Brewster, Arun, Long, Do, Hanash and Fahrmann.)

Purpose: The purpose of this study was to evaluate a deep-learning model (DLM) for classifying coronary arteries on coronary computed tomography -angiography (CCTA) using the Coronary Artery Disease-Reporting and Data System (CAD-RADS)., Materials and Methods: The DLM was trained with 10,800 curved multiplanar reformatted (cMPR) CCTA images classified by an expert radiologist using the CAD-RADS. For each of the three main coronary arteries, nine cMPR images 40° apart acquired around each arterial circumference were then classified by the DLM using the highest probability. For the validation set composed of 159 arteries from 53 consecutive patients, the images were read by two senior and two junior readers; consensus of the two seniors was the reference standard. With the DLM, the majority vote for the nine images was used to classify each artery. Three groups (CAD-RADS 0, 1-2, or 3-4-5) and 2 groups CAD-RADS 0-1-2 or 3-4-5 (<50% vs. ≥50% stenosis) were used for comparisons with readers and consensus. Performance of the model and readers was compared to the consensus reading using the intraclass coefficient (ICC) and Cohen's kappa coefficient at the artery and patient levels., Results: With the three groups at the artery level, the ICC of the DLM was 0.82 (95% CI: 0.75-0.88) and not significantly different from those of 3/4 readers; accuracy was 81%. With the binary classification, Cohen kappa coefficient of the DLM was 0.85 (95% CI: 0.69-0.94) and not significantly different from that of 3/4 readers; accuracy was 96%. At the patient level, sensitivity and specificity were 93% and 97% respectively, and the negative predictive value was 97%., Conclusion: The DLM detected ≥50% stenoses with performances similar to those achieved by senior radiologists., Competing Interests: Declaration of Competing Interest Jean-François Paul is founder of Spimed-AI Henri Giroussens is employed at Spimed-AI, (Copyright © 2022 Société française de radiologie. Published by Elsevier Masson SAS. All rights reserved.)

With the advancement in technology, machine learning can be applied to diagnose the mass/tumor in the brain using magnetic resonance imaging (MRI). This work proposes a novel developed transfer deep-learning model for the early diagnosis of brain tumors into their subclasses, such as pituitary, meningioma, and glioma. First, various layers of isolated convolutional-neural-network (CNN) models are built from scratch to check their performances for brain MRI images. Then, the 22-layer, binary-classification (tumor or no tumor) isolated-CNN model is re-utilized to re-adjust the neurons' weights for classifying brain MRI images into tumor subclasses using the transfer-learning concept. As a result, the developed transfer-learned model has a high accuracy of 95.75% for the MRI images of the same MRI machine. Furthermore, the developed transfer-learned model has also been tested using the brain MRI images of another machine to validate its adaptability, general capability, and reliability for real-time application in the future. The results showed that the proposed model has a high accuracy of 96.89% for an unseen brain MRI dataset. Thus, the proposed deep-learning framework can help doctors and radiologists diagnose brain tumors early. [ABSTRACT FROM AUTHOR]