Search

Background: Ensuring the efficient recognition and management of crop pests is crucial for maintaining the balance in global agricultural ecosystems and ecological harmony. Deep learning-based methods have shown promise in crop pest recognition. However, prevailing methods often fail to address a critical issue: biased pest training dataset distribution stemming from the tendency to collect images primarily in certain environmental contexts, such as paddy fields. This oversight hampers recognition accuracy when encountering pest images dissimilar to training samples, highlighting the need for a novel approach to overcome this limitation., Results: We introduce the Decoupled Feature Learning (DFL) framework, leveraging causal inference techniques to handle training dataset bias. DFL manipulates the training data based on classification confidence to construct different training domains and employs center triplet loss for learning class-core features. The proposed DFL framework significantly boosts existing baseline models, attaining unprecedented recognition accuracies of 95.33%, 92.59%, and 74.86% on the Li, DFSPD, and IP102 datasets, respectively., Conclusion: Extensive testing on three pest datasets using standard baseline models demonstrates the superiority of DFL in pest recognition. The visualization results show that DFL encourages the baseline models to capture the class-core features. The proposed DFL marks a pivotal step in mitigating the issue of data distribution bias, enhancing the reliability of deep learning in agriculture. © 2024 Society of Chemical Industry., (© 2024 Society of Chemical Industry.)

Rationale and Objectives: The aim of this study was to develop a deep learning radiomics nomogram (DLRN) based on B-mode ultrasound (BMUS) and color doppler flow imaging (CDFI) images for preoperative assessment of lymphovascular invasion (LVI) status in invasive breast cancer (IBC)., Materials and Methods: In this multicenter, retrospective study, 832 pathologically confirmed IBC patients were recruited from eight hospitals. The samples were divided into training, internal test, and external test sets. Deep learning and handcrafted radiomics features reflecting tumor phenotypes on BMUS and CDFI images were extracted. The BMUS score and CDFI score were calculated after radiomics feature selection. Subsequently, a DLRN was developed based on the scores and independent clinic-ultrasonic risk variables. The performance of the DLRN was evaluated for calibration, discrimination, and clinical usefulness., Results: The DLRN predicted the LVI with accuracy, achieving an area under the receiver operating characteristic curve of 0.93 (95% CI 0.90-0.95), 0.91 (95% CI 0.87-0.95), and 0.91 (95% CI 0.86-0.94) in the training, internal test, and external test sets, respectively, with good calibration. The DLRN demonstrated superior performance compared to the clinical model and single scores across all three sets (p < 0.05). Decision curve analysis and clinical impact curve confirmed the clinical utility of the model. Furthermore, significant enhancements in net reclassification improvement (NRI) and integrated discrimination improvement (IDI) indicated that the two scores could serve as highly valuable biomarkers for assessing LVI., Conclusion: The DLRN exhibited strong predictive value for LVI in IBC, providing valuable information for individualized treatment decisions., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved.)

Deep learning technology has been successfully used in Chest X-ray (CXR) images of COVID-19 patients. However, due to the characteristics of COVID-19 pneumonia and X-ray imaging, the deep learning methods still face many challenges, such as lower imaging quality, fewer training samples, complex radiological features and irregular shapes. To address these challenges, this study first introduces an extensive NSNP-like neuron model, and then proposes a multitask adversarial network architecture based on ENSNP-like neurons for chest X-ray images of COVID-19, called MAE-Net. The MAE-Net serves two tasks: (i) converting low-quality CXR images to high-quality images; (ii) classifying CXR images of COVID-19. The adversarial architecture of MAE-Net uses two generators and two discriminators, and two new loss functions have been introduced to guide the optimization of the network. The MAE-Net is tested on four benchmark COVID-19 CXR image datasets and compared them with eight deep learning models. The experimental results show that the proposed MAE-Net can enhance the conversion quality and the accuracy of image classification results.

Endoscopy is the primary modality for detecting asymptomatic esophageal squamous cell carcinoma (ESCC) and precancerous lesions. Improving detection rate remains challenging. We developed a system based on deep convolutional neural networks (CNNs) for detecting esophageal cancer and precancerous lesions [high-risk esophageal lesions (HrELs)] and validated its efficacy in improving HrEL detection rate in clinical practice (trial registration ChiCTR2100044126 at www.chictr.org.cn). Between April 2021 and March 2022, 3117 patients ≥50 years old were consecutively recruited from Taizhou Hospital, Zhejiang Province, and randomly assigned 1:1 to an experimental group (CNN-assisted endoscopy) or a control group (unassisted endoscopy) based on block randomization. The primary endpoint was the HrEL detection rate. In the intention-to-treat population, the HrEL detection rate [28 of 1556 (1.8%)] was significantly higher in the experimental group than in the control group [14 of 1561 (0.9%), P = 0.029], and the experimental group detection rate was twice that of the control group. Similar findings were observed between the experimental and control groups [28 of 1524 (1.9%) versus 13 of 1534 (0.9%), respectively; P = 0.021]. The system's sensitivity, specificity, and accuracy for detecting HrELs were 89.7, 98.5, and 98.2%, respectively. No adverse events occurred. The proposed system thus improved HrEL detection rate during endoscopy and was safe. Deep learning assistance may enhance early diagnosis and treatment of esophageal cancer and may become a useful tool for esophageal cancer screening.

Motivation: Accurate prediction of RNA subcellular localization plays an important role in understanding cellular processes and functions. Although post-transcriptional processes are governed by trans-acting RNA binding proteins (RBPs) through interaction with cis-regulatory RNA motifs, current methods do not incorporate RBP-binding information., Results: In this article, we propose DeepLocRNA, an interpretable deep-learning model that leverages a pre-trained multi-task RBP-binding prediction model to predict the subcellular localization of RNA molecules via fine-tuning. We constructed DeepLocRNA using a comprehensive dataset with variant RNA types and evaluated it on the held-out dataset. Our model achieved state-of-the-art performance in predicting RNA subcellular localization in mRNA and miRNA. It has also demonstrated great generalization capabilities, performing well on both human and mouse RNA. Additionally, a motif analysis was performed to enhance the interpretability of the model, highlighting signal factors that contributed to the predictions. The proposed model provides general and powerful prediction abilities for different RNA types and species, offering valuable insights into the localization patterns of RNA molecules and contributing to our understanding of cellular processes at the molecular level. A user-friendly web server is available at: https://biolib.com/KU/DeepLocRNA/., (© The Author(s) 2024. Published by Oxford University Press.)

Background: Deep learning provides an efficient automatic image recognition method for small bowel (SB) capsule endoscopy (CE) that can assist physicians in diagnosis. However, the existing deep learning models present some unresolved challenges., Aim: To propose a novel and effective classification and detection model to automatically identify various SB lesions and their bleeding risks, and label the lesions accurately so as to enhance the diagnostic efficiency of physicians and the ability to identify high-risk bleeding groups., Methods: The proposed model represents a two-stage method that combined image classification with object detection. First, we utilized the improved ResNet-50 classification model to classify endoscopic images into SB lesion images, normal SB mucosa images, and invalid images. Then, the improved YOLO-V5 detection model was utilized to detect the type of lesion and its risk of bleeding, and the location of the lesion was marked. We constructed training and testing sets and compared model-assisted reading with physician reading., Results: The accuracy of the model constructed in this study reached 98.96%, which was higher than the accuracy of other systems using only a single module. The sensitivity, specificity, and accuracy of the model-assisted reading detection of all images were 99.17%, 99.92%, and 99.86%, which were significantly higher than those of the endoscopists' diagnoses. The image processing time of the model was 48 ms/image, and the image processing time of the physicians was 0.40 ± 0.24 s/image ( P < 0.001)., Conclusion: The deep learning model of image classification combined with object detection exhibits a satisfactory diagnostic effect on a variety of SB lesions and their bleeding risks in CE images, which enhances the diagnostic efficiency of physicians and improves the ability of physicians to identify high-risk bleeding groups., Competing Interests: Conflict-of-interest statement: The authors declare having no conflicts of interest., (©The Author(s) 2024. Published by Baishideng Publishing Group Inc. All rights reserved.)

Cancer metastasis is one of the main causes of cancer progression and difficulty in treatment. Genes play a key role in the process of cancer metastasis, as they can influence tumor cell invasiveness, migration ability and fitness. At the same time, there is heterogeneity in the organs of cancer metastasis. Breast cancer, prostate cancer, etc. tend to metastasize in the bone. Previous studies have pointed out that the occurrence of metastasis is closely related to which tissue is transferred to and genes. In this paper, we identified genes associated with cancer metastasis to different tissues based on LASSO and Pearson correlation coefficients. In total, we identified 45 genes associated with bone metastases, 89 genes associated with lung metastases, and 86 genes associated with liver metastases. Through the expression of these genes, we propose a CNN-based model to predict the occurrence of metastasis. We call this method MDCNN, which introduces a modulation mechanism that allows the weights of convolution kernels to be adjusted at different positions and feature maps, thereby adaptively changing the convolution operation at different positions. Experiments have proved that MDCNN has achieved satisfactory prediction accuracy in bone metastasis, lung metastasis and liver metastasis, and is better than other 4 methods of the same kind. We performed enrichment analysis and immune infiltration analysis on bone metastasis-related genes, and found multiple pathways and GO terms related to bone metastasis, and found that the abundance of macrophages and monocytes was the highest in patients with bone metastasis., Competing Interests: Declaration of competing interest The authors declare that they have no competing interests., (Copyright © 2023. Published by Elsevier Ltd.)

Background: Long-term improvements in air quality and public health in the continental USA were disrupted over the past decade by increased fire emissions that potentially offset the decrease in anthropogenic emissions. This study aims to estimate trends in black carbon and PM 2·5 concentrations and their attributable mortality burden across the USA., Methods: In this study, we derived daily concentrations of PM 2·5 and its highly toxic black carbon component at a 1-km resolution in the USA from 2000 to 2020 via deep learning that integrated big data from satellites, models, and surface observations. We estimated the annual PM 2·5 -attributable and black carbon-attributable mortality burden at each 1-km 2 grid using concentration-response functions collected from a national cohort study and a meta-analysis study, respectively. We investigated the spatiotemporal linear-regressed trends in PM 2·5 and black carbon pollution and their associated premature deaths from 2000 to 2020, and the impact of wildfires on air quality and public health., Findings: Our results showed that PM 2·5 and black carbon estimates are reliable, with sample-based cross-validated coefficients of determination of 0·82 and 0·80, respectively, for daily estimates (0·97 and 0·95 for monthly estimates). Both PM 2·5 and black carbon in the USA showed significantly decreasing trends overall during 2000 to 2020 (22% decrease for PM 2·5 and 11% decrease for black carbon), leading to a reduction of around 4200 premature deaths per year (95% CI 2960-5050). However, since 2010, the decreasing trends of fine particles and premature deaths have reversed to increase in the western USA (55% increase in PM 2·5 , 86% increase in black carbon, and increase of 670 premature deaths [460-810]), while remaining mostly unchanged in the eastern USA. The western USA showed large interannual fluctuations that were attributable to the increasing incidence of wildfires. Furthermore, the black carbon-to-PM 2·5 mass ratio increased annually by 2·4% across the USA, mainly due to increasing wildfire emissions in the western USA and more rapid reductions of other components in the eastern USA, suggesting a potential increase in the relative toxicity of PM 2·5 . 100% of populated areas in the USA have experienced at least one day of PM 2·5 pollution exceeding the daily air quality guideline level of 15 μg/m 3 during 2000-2020, with 99% experiencing at least 7 days and 85% experiencing at least 30 days. The recent widespread wildfires have greatly increased the daily exposure risks in the western USA, and have also impacted the midwestern USA due to the long-range transport of smoke., Interpretation: Wildfires have become increasingly intensive and frequent in the western USA, resulting in a significant increase in smoke-related emissions in populated areas. This increase is likely to have contributed to a decline in air quality and an increase in attributable mortality. Reducing fire risk via effective policies besides mitigation of climate warming, such as wildfire prevention and management, forest restoration, and new revenue generation, could substantially improve air quality and public health in the coming decades., Funding: National Aeronautics and Space Administration (NASA) Applied Science programme, NASA MODIS maintenance programme, NASA MAIA satellite mission programme, NASA GMAO core fund, National Oceanic and Atmospheric Administration (NOAA) GEO-XO project, NOAA Atmospheric Chemistry, Carbon Cycle, and Climate (AC4) programme, and NOAA Educational Partnership Program with Minority Serving Institutions., Competing Interests: Declaration of interests We declare no competing interests., (Copyright © 2023 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Published by Elsevier Ltd.. All rights reserved.)

Fine particulate matter (PM 2.5 ) chemical composition has strong and diverse impacts on the planetary environment, climate, and health. These effects are still not well understood due to limited surface observations and uncertainties in chemical model simulations. We developed a four-dimensional spatiotemporal deep forest (4D-STDF) model to estimate daily PM 2.5 chemical composition at a spatial resolution of 1 km in China since 2000 by integrating measurements of PM 2.5 species from a high-density observation network, satellite PM 2.5 retrievals, atmospheric reanalyses, and model simulations. Cross-validation results illustrate the reliability of sulfate (SO 4 2- ), nitrate (NO 3 - ), ammonium (NH 4 + ), and chloride (Cl - ) estimates, with high coefficients of determination (CV-R 2 ) with ground-based observations of 0.74, 0.75, 0.71, and 0.66, and average root-mean-square errors (RMSE) of 6.0, 6.6, 4.3, and 2.3 μg/m 3 , respectively. The three components of secondary inorganic aerosols (SIAs) account for 21% (SO 4 2- ), 20% (NO 3 - ), and 14% (NH 4 + ) of the total PM 2.5 mass in eastern China; we observed significant reductions in the mass of inorganic components by 40-43% between 2013 and 2020, slowing down since 2018. Comparatively, the ratio of SIA to PM 2.5 increased by 7% across eastern China except in Beijing and nearby areas, accelerating in recent years. SO 4 2- has been the dominant SIA component in eastern China, although it was surpassed by NO 3 - in some areas, e.g., Beijing-Tianjin-Hebei region since 2016. SIA, accounting for nearly half (∼46%) of the PM 2.5 mass, drove the explosive formation of winter haze episodes in the North China Plain. A sharp decline in SIA concentrations and an increase in SIA-to-PM 2.5 ratios during the COVID-19 lockdown were also revealed, reflecting the enhanced atmospheric oxidation capacity and formation of secondary particles.

Adverse Drug Reactions (ADRs) have a direct impact on human health. As continuous pharmacovigilance and drug monitoring prove to be costly and time-consuming, computational methods have emerged as promising alternatives. However, most existing computational methods primarily focus on predicting whether or not the drug is associated with an adverse reaction and do not consider the core issue of drug benefit-risk assessment-whether the treatment outcome is serious when adverse drug reactions occur. To this end, we categorize serious clinical outcomes caused by adverse reactions to drugs into seven distinct classes and present a deep learning framework, so-called GCAP, for predicting the seriousness of clinical outcomes of adverse reactions to drugs. GCAP has two tasks: one is to predict whether adverse reactions to drugs cause serious clinical outcomes, and the other is to infer the corresponding classes of serious clinical outcomes. Experimental results demonstrate that our method is a powerful and robust framework with high extendibility. GCAP can serve as a useful tool to successfully address the challenge of predicting the seriousness of clinical outcomes stemming from adverse reactions to drugs., (© 2023. Springer Nature Limited.)

Flexible wearable pressure sensors have received increasing attention as the potential application of flexible wearable devices in human health monitoring and artificial intelligence. However, the complex and expensive process of the conductive filler has limited its practical production and application on a large scale to a certain extent. This study presents a kind of piezoresistive sensor by sinking nonwoven fabrics (NWFs) into tungsten disulfide (WS 2 ) and Ti 3 C 2 T x MXene solutions. With the advantages of a simple production process and practicality, it is conducive to the realization of large-scale production. The assembled flexible pressure sensor exhibits high sensitivity (45.81 kPa -1 ), wide detection range (0-410 kPa), fast response/recovery time (18/36 ms), and excellent stability and long-term durability (up to 5000 test cycles). Because of the high elastic modulus of MXene and the synergistic effect between WS 2 and MXene, the detection range and sensitivity of the piezoresistive pressure sensor are greatly improved, realizing the stable detection of human motion status in all directions. Meanwhile, its high sensitivity at low pressure allows the sensor to accurately detect weak signals such as weak airflow and wrist pulses. In addition, combining the sensor with deep-learning makes it easy to recognize human respiratory patterns with high accuracy, demonstrating its potential impact in the fields of ergonomics and low-cost flexible electronics.

Gene mutation detection is usually carried out by molecular biological methods, which is expensive and has a long-time cycle. In contrast, pathological images are ubiquitous. If clinically significant gene mutations can be predicted only through pathological images, it will greatly promote the widespread use of gene mutation detection in clinical practice. However, current gene mutation prediction methods based on pathological images are ineffective because of the inability to identify mutated regions in gigapixel Whole Slide Image (WSI). To address this challenge, hereby we propose a carefully designed framework for WSI-based gene mutation prediction, which consists of three parts. (i) The first part of cancerous area segmentation, based on supervised learning, quickly filters out a large number of non-mutated regions; (ii) the second part of cancerous patch clustering, based on the representations derived from contrastive learning, ensures the comprehensiveness of patch selection; and (iii) the third part of mutation classification, based on the proposed hierarchical deep multi-instance learning method (HDMIL), ensures that sufficient patches are considered and inaccurate selections are ignored. In addition, benefiting from a two-stage attention mechanism in HDMIL, the patches that are highly correlated with gene mutations can be identified. This interpretability can help a pathologist to analyze the correlation between gene mutation and histopathological morphology. Experimental results demonstrate that the proposed gene mutation prediction framework significantly outperforms the state-of-the-art methods. In the TCGA bladder cancer dataset, five clinically relevant gene mutations are well predicted., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2023 Elsevier B.V. All rights reserved.)

Background: Preoperative planning of orthognathic surgery is indispensable for achieving ideal surgical outcome regarding the occlusion and jaws' position. However, orthognathic surgery planning is sophisticated and highly experience-dependent, which requires comprehensive consideration of facial morphology and occlusal function. This study aimed to investigate a robust and automatic method based on deep learning to predict reposition vectors of jawbones in orthognathic surgery plan., Methods: A regression neural network named VSP transformer was developed based on Transformer architecture. Firstly, 3D cephalometric analysis was employed to quantify skeletal-facial morphology as input features. Next, input features were weighted using pretrained results to minimize bias resulted from multicollinearity. Through encoder-decoder blocks, ten landmark-based reposition vectors of jawbones were predicted. Permutation importance (PI) method was used to calculate contributions of each feature to final prediction to reveal interpretability of the proposed model., Results: VSP transformer model was developed with 383 samples and clinically tested with 49 prospectively collected samples. Our proposed model outperformed other four classic regression models in prediction accuracy. Mean absolute errors (MAE) of prediction were 1.41 mm in validation set and 1.34 mm in clinical test set. The interpretability results of the model were highly consistent with clinical knowledge and experience., Conclusions: The developed model can predict reposition vectors of orthognathic surgery plan with high accuracy and good clinically practical-effectiveness. Moreover, the model was proved reliable because of its good interpretability., (© 2023. The Author(s).)

Deep learning-based in silico alternatives have been demonstrated to be of significant importance in the acceleration of the drug discovery process and enhancement of success rates. Cyclin-dependent kinase 12 (CDK12) is a transcription-related cyclin-dependent kinase that may act as a biomarker and therapeutic target for cancers. However, currently, there is no high selective CDK12 inhibitor in clinical development and the identification of new specific CDK12 inhibitors has become increasingly challenging due to their similarity with CDK13. In this study, we developed a virtual screening workflow that combines deep learning with virtual screening tools and can be applied rapidly to millions of molecules. We designed a Transformer architecture Drug-Target Interaction (DTI) model with dual-branched self-supervised pre-trained molecular graph models and protein sequence models. Our predictive model produced satisfactory predictions for various targets, including CDK12, with several novel hits. We screened a large compound library consisting of 4.5 million drug-like molecules and recommended a list of potential CDK12 inhibitors for further experimental testing. In kinase assay, compared to the positive CDK12 inhibitor THZ531, the compounds CICAMPA-01, 02, 03 displayed more effective inhibition of CDK12, up to three times as much as THZ531. The compounds CICAMPA-03, 05, 04, 07 showed less inhibition of CDK13 compare to THZ531. In vitro, the IC50 of CICAMPA-01, 04, 05, 06, 09 was less than 3 μM in the HER2 positive CDK12 amplification breast cancer cell line BT-474. Overall, this study provides a highly efficient and end-to-end deep learning protocol, in conjunction with molecular docking, for discovering CDK12 inhibitors in cancers. Additionally, we disclose five novel CDK12 inhibitors. These results may accelerate the discovery of novel chemical-class drugs for cancer treatment., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2023 The Authors. Published by Elsevier Masson SAS.. All rights reserved.)

Purpose: Many deep learning methods have been developed for pulmonary lesion detection in chest computed tomography (CT) images. However, these methods generally target one particular lesion type, that is, pulmonary nodules. In this work, we intend to develop and evaluate a novel deep learning method for a more challenging task, detecting various benign and malignant mediastinal lesions with wide variations in sizes, shapes, intensities, and locations in chest CT images., Methods: Our method for mediastinal lesion detection contains two main stages: (a) size-adaptive lesion candidate detection followed by (b) false-positive (FP) reduction and benign-malignant classification. For candidate detection, an anchor-free and one-stage detector, namely 3D-CenterNet is designed to locate suspicious regions (i.e., candidates with various sizes) within the mediastinum. Then, a 3D-SEResNet-based classifier is used to differentiate FPs, benign lesions, and malignant lesions from the candidates., Results: We evaluate the proposed method by conducting five-fold cross-validation on a relatively large-scale dataset, which consists of data collected on 1136 patients from a grade A tertiary hospital. The method can achieve sensitivity scores of 84.3% ± 1.9%, 90.2% ± 1.4%, 93.2% ± 0.8%, and 93.9% ± 1.1%, respectively, in finding all benign and malignant lesions at 1/8, 1/4, ½, and 1 FPs per scan, and the accuracy of benign-malignant classification can reach up to 78.7% ± 2.5%., Conclusions: The proposed method can effectively detect mediastinal lesions with various sizes, shapes, and locations in chest CT images. It can be integrated into most existing pulmonary lesion detection systems to promote their clinical applications. The method can also be readily extended to other similar 3D lesion detection tasks., (© 2022 American Association of Physicists in Medicine.)

The persistence of the global COVID-19 pandemic caused by the SARS-CoV-2 virus has continued to emphasize the need for point-of-care (POC) diagnostic tests for viral diagnosis. The most widely used tests, lateral flow assays used in rapid antigen tests, and reverse-transcriptase real-time polymerase chain reaction (RT-PCR), have been instrumental in mitigating the impact of new waves of the pandemic, but fail to provide both sensitive and rapid readout to patients. Here, we present a portable lens-free imaging system coupled with a particle agglutination assay as a novel biosensor for SARS-CoV-2. This sensor images and quantifies individual microbeads undergoing agglutination through a combination of computational imaging and deep learning as a way to detect levels of SARS-CoV-2 in a complex sample. SARS-CoV-2 pseudovirus in solution is incubated with acetyl cholinesterase 2 (ACE2)-functionalized microbeads then loaded into an inexpensive imaging chip. The sample is imaged in a portable in-line lens-free holographic microscope and an image is reconstructed from a pixel superresolved hologram. Images are analyzed by a deep-learning algorithm that distinguishes microbead agglutination from cell debris and viral particle aggregates, and agglutination is quantified based on the network output. We propose an assay procedure using two images which results in the accurate determination of viral concentrations greater than the limit of detection (LOD) of 1.27 × 10 3 copies per mL, with a tested dynamic range of 3 orders of magnitude, without yet reaching the upper limit. This biosensor can be used for fast SARS-CoV-2 diagnosis in low-resource POC settings and has the potential to mitigate the spread of future waves of the pandemic.

Purpose: To apply deep learning (DL) techniques to develop an automatic intelligent classification system identifying the specific types of myopic maculopathy (MM) based on macular optical coherence tomography (OCT) images using transfer learning (TL)., Method: In this retrospective study, a total of 3,945 macular OCT images from 2,866 myopic patients were recruited from the ophthalmic outpatients of three hospitals. After culling out 545 images with poor quality, a dataset containing 3,400 macular OCT images was manually classified according to the ATN system, containing four types of MM with high OCT diagnostic values. Two DL classification algorithms were trained to identify the targeted lesion categories: Algorithm A was trained from scratch, and algorithm B using the TL approach initiated from the classification algorithm developed in our previous study. After comparing the training process, the algorithm with better performance was tested and validated. The performance of the classification algorithm in the test and validation sets was evaluated using metrics including sensitivity, specificity, accuracy, quadratic-weighted kappa score, and the area under the receiver operating characteristic curve (AUC). Moreover, the human-machine comparison was conducted. To better evaluate the algorithm and clarify the optimization direction, the dimensionality reduction analysis and heat map analysis were also used to visually analyze the algorithm., Results: Algorithm B showed better performance in the training process. In the test set, the algorithm B achieved relatively robust performance with macro AUC, accuracy, and quadratic-weighted kappa of 0.986, 96.04% (95% CI: 0.951, 0.969), and 0.940 (95% CI: 0.909-0.971), respectively. In the external validation set, the performance of algorithm B was slightly inferior to that in the test set. In human-machine comparison test, the algorithm indicators were inferior to the retinal specialists but were the same as the ordinary ophthalmologists. In addition, dimensionality reduction visualization and heatmap visualization analysis showed excellent performance of the algorithm., Conclusion: Our macular OCT image classification algorithm developed using the TL approach exhibited excellent performance. The automatic diagnosis system for macular OCT images of MM based on DL showed potential application prospects., Competing Interests: The 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 He, Ren, Lu, Tang, Wang, Yang and Han.)

The human gut microbiome encodes a large variety of antimicrobial peptides (AMPs), but the short lengths of AMPs pose a challenge for computational prediction. Here we combined multiple natural language processing neural network models, including LSTM, Attention and BERT, to form a unified pipeline for candidate AMP identification from human gut microbiome data. Of 2,349 sequences identified as candidate AMPs, 216 were chemically synthesized, with 181 showing antimicrobial activity (a positive rate of >83%). Most of these peptides have less than 40% sequence homology to AMPs in the training set. Further characterization of the 11 most potent AMPs showed high efficacy against antibiotic-resistant, Gram-negative pathogens and demonstrated significant efficacy in lowering bacterial load by more than tenfold against a mouse model of bacterial lung infection. Our study showcases the potential of machine learning approaches for mining functional peptides from metagenome data and accelerating the discovery of promising AMP candidate molecules for in-depth investigations., (© 2022. The Author(s), under exclusive licence to Springer Nature America, Inc.)

Objectives/hypothesis: To develop a deep-learning-based automatic diagnosis system for identifying nasopharyngeal carcinoma (NPC) from noncancer (inflammation and hyperplasia), using both white light imaging (WLI) and narrow-band imaging (NBI) nasopharyngoscopy images., Study Design: Retrospective study., Methods: A total of 4,783 nasopharyngoscopy images (2,898 WLI and 1,885 NBI) of 671 patients were collected and a novel deep convolutional neural network (DCNN) framework was developed named Siamese deep convolutional neural network (S-DCNN), which can simultaneously utilize WLI and NBI images to improve the classification performance. To verify the effectiveness of combining the above-mentioned two modal images for prediction, we compared the proposed S-DCNN with two baseline models, namely DCNN-1 (only considering WLI images) and DCNN-2 (only considering NBI images)., Results: In the threefold cross-validation, an overall accuracy and area under the curve of the three DCNNs achieved 94.9% (95% confidence interval [CI] 93.3%-96.5%) and 0.986 (95% CI 0.982-0.992), 87.0% (95% CI 84.2%-89.7%) and 0.930 (95% CI 0.906-0.961), and 92.8% (95% CI 90.4%-95.3%) and 0.971 (95% CI 0.953-0.992), respectively. The accuracy of S-DCNN is significantly improved compared with DCNN-1 (P-value <.001) and DCNN-2 (P-value = .008)., Conclusion: Using the deep-learning technology to automatically diagnose NPC under nasopharyngoscopy can provide valuable reference for NPC screening. Superior performance can be obtained by simultaneously utilizing the multimodal features of NBI image and WLI image of the same patient., Level of Evidence: 3 Laryngoscope, 132:999-1007, 2022., (© 2021 The American Laryngological, Rhinological and Otological Society, Inc.)

Purpose: This study aimed to determine the efficacy of a multimodal deep learning (DL) model using optical coherence tomography (OCT) and optical coherence tomography angiography (OCTA) images for the assessment of choroidal neovascularization (CNV) in neovascular age-related macular degeneration (AMD)., Methods: This retrospective and cross-sectional study was performed at a multicentre, and the inclusion criteria were age >50 years and a diagnosis of typical neovascular AMD. The OCT and OCTA data for an internal data set and two external data sets were collected. A DL model was developed with a novel feature-level fusion (FLF) method utilized to combine the multimodal data. The results were compared with identification performed by an ophthalmologist. The best model was tested on two external data sets to show its potential for clinical use., Results: Our best model achieved an accuracy of 95.5% and an area under the curve (AUC) of 0.9796 on multimodal data inputs for the internal data set, which is comparable to the performance of retinal specialists. The proposed model reached an accuracy of 100.00% and an AUC of 1.0 for the Ningbo data set, and these performance indicators were 90.48% and an AUC of 0.9727 for the Jinhua data set., Conclusion: The FLF method is feasible and highly accurate, and could enhance the power of the existing computer-aided diagnosis systems. The bi-modal computer-aided diagnosis (CADx) system for the automated identification of CNV activity is an accurate and promising tool in the realm of public health., (© 2021 Acta Ophthalmologica Scandinavica Foundation. Published by John Wiley & Sons Ltd.)

Purpose: Pneumothorax is a life-threatening emergency that requires immediate treatment. Frontal-view chest X-ray images are typically used for pneumothorax detection in clinical practice. However, manual review of radiographs is time-consuming, labor-intensive, and highly dependent on the experience of radiologists, which may lead to misdiagnosis. Here, we aim to develop a reliable automatic classification method to assist radiologists in rapidly and accurately diagnosing pneumothorax in frontal chest radiographs., Methods: A novel residual neural network (ResNet)-based two-stage deep-learning strategy is proposed for pneumothorax identification: local feature learning (LFL) followed by global multi-instance learning (GMIL). Most of the nonlesion regions in the images are removed for learning discriminative features. Two datasets are used for large-scale validation: a private dataset (27 955 frontal-view chest X-ray images) and a public dataset (the National Institutes of Health [NIH] ChestX-ray14; 112 120 frontal-view X-ray images). The model performance of the identification was evaluated using the accuracy, precision, recall, specificity, F1-score, receiver operating characteristic (ROC), and area under ROC curve (AUC). Fivefold cross-validation is conducted on the datasets, and then the mean and standard deviation of the above-mentioned metrics are calculated to assess the overall performance of the model., Results: The experimental results demonstrate that the proposed learning strategy can achieve state-of-the-art performance on the NIH dataset with an accuracy, AUC, precision, recall, specificity, and F1-score of 94.4% ± 0.7%, 97.3% ± 0.5%, 94.2% ± 0.3%, 94.6% ± 1.5%, 94.2% ± 0.4%, and 94.4% ± 0.7%, respectively., Conclusions: The experimental results demonstrate that our proposed CAD system is an efficient assistive tool in the identification of pneumothorax., (© 2021 American Association of Physicists in Medicine.)

An important reason of cancer proliferation is the change in DNA methylation patterns, characterized by the localized hypermethylation of the promoters of tumor-suppressor genes together with an overall decrease in the level of 5-methylcytosine (5mC). Therefore, identifying the 5mC sites in the promoters is a critical step towards further understanding the diverse functions of DNA methylation in genetic diseases such as cancers and aging. However, most wet-lab experimental techniques are often time consuming and laborious for detecting 5mC sites. In this study, we proposed a deep learning-based approach, called BiLSTM-5mC, for accurately identifying 5mC sites in genome-wide DNA promoters. First, we randomly divided the negative samples into 11 subsets of equal size, one of which can form the balance subset by combining with the positive samples in the same amount. Then, two types of feature vectors encoded by the one-hot method, and the nucleotide property and frequency (NPF) methods were fed into a bidirectional long short-term memory (BiLSTM) network and a full connection layer to train the 22 submodels. Finally, the outputs of these models were integrated to predict 5mC sites by using the majority vote strategy. Our experimental results demonstrated that BiLSTM-5mC outperformed existing methods based on the same independent dataset.

The accumulation of protein sequence and structure data allows researchers to obtain large amount of descriptive information, simultaneously it poses an urgent need for researchers to extract information from existing data efficiently and apply it to downstream tasks. Protein design enables the development of novel proteins that are no longer restricted by experimental conditions, which is of great significance for drug target prediction, drug discovery, and material design. As an efficient method for data feature extraction, deep learning can be used to model protein data, and further add a priori information to design novel proteins. Therefore, protein design based on deep learning has become a promising approach despite of many challenges. This review summarizes the deep learning-based modeling and design methods of protein sequence and structure data, highlighting the strategies, principle, scope of application and case studies, with the aim to provide a valuable reference for relevant researchers.

Purpose: The purpose of this study was to engineer deep learning (DL) models that can identify myopic maculopathy in patients with high myopia based on optical coherence tomography (OCT) images., Methods: An artificial intelligence (AI) system was developed using 2342 qualified OCT macular images from 1041 patients with pathologic myopia admitted to the Affiliated Eye Hospital of Wenzhou Medical University (WMU). We adopted an ResNeSt101 architecture to train five independent models to identify the following five myopic maculopathies: macular choroidal thinning, macular Bruch membrane (BM) defects, subretinal hyper-reflective material (SHRM), myopic traction maculopathy (MTM), and dome-shaped macula (DSM). We tested the models with an independent test dataset that included 450 images obtained from 297 patients with high myopia. Focal loss was used to address class imbalance, and optimal operating thresholds were determined according to the Youden Index. The performance was quantified using the area under the receiver operating characteristic (AUC), sensitivity, specificity, and confusion matrix., Results: For the identification of myopic maculopathy, the AUCs of receiver operating characteristic (ROC) curves were 0.927 to 0.974 for 5 myopic maculopathies. Our AI system achieved sensitivities equal to or even better than those of junior retinal specialists (56.16-99.73%). The diagnosis of it is also interpretable that we provide visual explanations clearly via heatmaps., Conclusions: We developed a convolutional neural network (CNN)-based DL AI system for detection and classification of myopic maculopathy in patients with high myopia using OCT macular images. Our AI system achieved sensitivities equal to or even better than those of junior retinal specialists., Translational Relevance: This AI system can be widely applied in sophisticated situations in large-scale high myopia screening.

Coronavirus disease 2019 (COVID-19) is one of the most destructive pandemic after millennium, forcing the world to tackle a health crisis. Automated lung infections classification using chest X-ray (CXR) images could strengthen diagnostic capability when handling COVID-19. However, classifying COVID-19 from pneumonia cases using CXR image is a difficult task because of shared spatial characteristics, high feature variation and contrast diversity between cases. Moreover, massive data collection is impractical for a newly emerged disease, which limited the performance of data thirsty deep learning models. To address these challenges, Multiscale Attention Guided deep network with Soft Distance regularization (MAG-SD) is proposed to automatically classify COVID-19 from pneumonia CXR images. In MAG-SD, MA-Net is used to produce prediction vector and attention from multiscale feature maps. To improve the robustness of trained model and relieve the shortage of training data, attention guided augmentations along with a soft distance regularization are posed, which aims at generating meaningful augmentations and reduce noise. Our multiscale attention model achieves better classification performance on our pneumonia CXR image dataset. Plentiful experiments are proposed for MAG-SD which demonstrates its unique advantage in pneumonia classification over cutting-edge models. The code is available at https://github.com/JasonLeeGHub/MAG-SD.

Objective: Computed tomography (CT) provides rich diagnosis and severity information of COVID-19 in clinical practice. However, there is no computerized tool to automatically delineate COVID-19 infection regions in chest CT scans for quantitative assessment in advanced applications such as severity prediction. The aim of this study was to develop a deep learning (DL)-based method for automatic segmentation and quantification of infection regions as well as the entire lungs from chest CT scans., Methods: The DL-based segmentation method employs the "VB-Net" neural network to segment COVID-19 infection regions in CT scans. The developed DL-based segmentation system is trained by CT scans from 249 COVID-19 patients, and further validated by CT scans from other 300 COVID-19 patients. To accelerate the manual delineation of CT scans for training, a human-involved-model-iterations (HIMI) strategy is also adopted to assist radiologists to refine automatic annotation of each training case. To evaluate the performance of the DL-based segmentation system, three metrics, that is, Dice similarity coefficient, the differences of volume, and percentage of infection (POI), are calculated between automatic and manual segmentations on the validation set. Then, a clinical study on severity prediction is reported based on the quantitative infection assessment., Results: The proposed DL-based segmentation system yielded Dice similarity coefficients of 91.6% ± 10.0% between automatic and manual segmentations, and a mean POI estimation error of 0.3% for the whole lung on the validation dataset. Moreover, compared with the cases with fully manual delineation that often takes hours, the proposed HIMI training strategy can dramatically reduce the delineation time to 4 min after three iterations of model updating. Besides, the best accuracy of severity prediction was 73.4% ± 1.3% when the mass of infection (MOI) of multiple lung lobes and bronchopulmonary segments were used as features for severity prediction, indicating the potential clinical application of our quantification technique on severity prediction., Conclusions: A DL-based segmentation system has been developed to automatically segment and quantify infection regions in CT scans of COVID-19 patients. Quantitative evaluation indicated high accuracy in automatic infection delineation and severity prediction., (© 2020 American Association of Physicists in Medicine.)

Electrocardiography (ECG) is essential in many heart diseases. However, some ECGs are recorded by paper, which can be highly noisy. Digitizing the paper-based ECG records into a high-quality signal is critical for further analysis. We formulated the digitization problem as a segmentation problem and proposed a deep learning method to digitize highly noisy ECG scans. Our method extracts the ECG signal in an end-to-end manner and can handle different paper record layouts. In the experiment, our model clearly extracted the ECG waveform with a Dice coefficient of 0.85 and accurately measured the common ECG parameters with more than 0.90 Pearson's correlation. We showed that the end-to-end approach with deep learning can be powerful in ECG digitization. To the best of our knowledge, we provide the first approach to digitize the least informative noisy binary ECG scans and potentially be generalized to digitize various ECG records., (Copyright © 2020 The Authors. Published by Elsevier Ltd.. All rights reserved.)

Precise biomarker development is a key step in disease management. However, most of the published biomarkers were derived from a relatively small number of samples with supervised approaches. Recent advances in unsupervised machine learning promise to leverage very large datasets for making better predictions of disease biomarkers. Denoising autoencoder (DA) is one of the unsupervised deep learning algorithms, which is a stochastic version of autoencoder techniques. The principle of DA is to force the hidden layer of autoencoder to capture more robust features by reconstructing a clean input from a corrupted one. Here, a DA model was applied to analyze integrated transcriptomic data from 13 published lung cancer studies, which consisted of 1916 human lung tissue samples. Using DA, we discovered a molecular signature composed of multiple genes for lung adenocarcinoma (ADC). In independent validation cohorts, the proposed molecular signature is proved to be an effective classifier for lung cancer histological subtypes. Also, this signature successfully predicts clinical outcome in lung ADC, which is independent of traditional prognostic factors. More importantly, this signature exhibits a superior prognostic power compared with the other published prognostic genes. Our study suggests that unsupervised learning is helpful for biomarker development in the era of precision medicine., Competing Interests: Competing interests The authors have declared no competing interests., (Copyright © 2020 The Authors. Published by Elsevier B.V. All rights reserved.)

We propose a conceptually simple framework for fast COVID-19 screening in 3D chest CT images. The framework can efficiently predict whether or not a CT scan contains pneumonia while simultaneously identifying pneumonia types between COVID-19 and Interstitial Lung Disease (ILD) caused by other viruses. In the proposed method, two 3D-ResNets are coupled together into a single model for the two above-mentioned tasks via a novel prior-attention strategy. We extend residual learning with the proposed prior-attention mechanism and design a new so-called prior-attention residual learning (PARL) block. The model can be easily built by stacking the PARL blocks and trained end-to-end using multi-task losses. More specifically, one 3D-ResNet branch is trained as a binary classifier using lung images with and without pneumonia so that it can highlight the lesion areas within the lungs. Simultaneously, inside the PARL blocks, prior-attention maps are generated from this branch and used to guide another branch to learn more discriminative representations for the pneumonia-type classification. Experimental results demonstrate that the proposed framework can significantly improve the performance of COVID-19 screening. Compared to other methods, it achieves a state-of-the-art result. Moreover, the proposed method can be easily extended to other similar clinical applications such as computer-aided detection and diagnosis of pulmonary nodules in CT images, glaucoma lesions in Retina fundus images, etc.

Purpose: In clinical practice, invasiveness is an important reference indicator for differentiating the malignant degree of subsolid pulmonary nodules. These nodules can be classified as atypical adenomatous hyperplasia (AAH), adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA), or invasive adenocarcinoma (IAC). The automatic determination of a nodule's invasiveness based on chest CT scans can guide treatment planning. However, it is challenging, owing to the insufficiency of training data and their interclass similarity and intraclass variation. To address these challenges, we propose a two-stage deep learning strategy for this task: prior-feature learning followed by adaptive-boost deep learning., Methods: The adaptive-boost deep learning is proposed to train a strong classifier for invasiveness classification of subsolid nodules in chest CT images, using multiple 3D convolutional neural network (CNN)-based weak classifiers. Because ensembles of multiple deep 3D CNN models have a huge number of parameters and require large computing resources along with more training and testing time, the prior-feature learning is proposed to reduce the computations by sharing the CNN layers between all weak classifiers. Using this strategy, all weak classifiers can be integrated into a single network., Results: Tenfold cross validation of binary classification was conducted on a total of 1357 nodules, including 765 noninvasive (AAH and AIS) and 592 invasive nodules (MIA and IAC). Ablation experimental results indicated that the proposed binary classifier achieved an accuracy of 73.4 \% ± 1.4 with an AUC of 81.3 \% ± 2.2 . These results are superior compared to those achieved by three experienced chest imaging specialists who achieved an accuracy of 69.1 \% , 69.3 \% , and 67.9 \% , respectively. About 200 additional nodules were also collected. These nodules covered 50 cases for each category (AAH, AIS, MIA, and IAC, respectively). Both binary and multiple classifications were performed on these data and the results demonstrated that the proposed method definitely achieves better performance than the performance achieved by nonensemble deep learning methods., Conclusions: It can be concluded that the proposed adaptive-boost deep learning can significantly improve the performance of invasiveness classification of pulmonary subsolid nodules in CT images, while the prior-feature learning can significantly reduce the total size of deep models. The promising results on clinical data show that the trained models can be used as an effective lung cancer screening tool in hospitals. Moreover, the proposed strategy can be easily extended to other similar classification tasks in 3D medical images., (© 2020 American Association of Physicists in Medicine.)

The CRISPR-Cas9 system has become the most promising and versatile tool for genetic manipulation applications. Albeit the technology has been broadly adopted by both academic and pharmaceutic societies, the activity (on-target) and specificity (off-target) of CRISPR-Cas9 are decisive factors for any application of the technology. Several in silico gRNA activity and specificity predicting models and web tools have been developed, making it much more convenient and precise for conducting CRISPR gene editing studies. In this review, we present an overview and comparative analysis of machine and deep learning (MDL)-based algorithms, which are believed to be the most effective and reliable methods for the prediction of CRISPR gRNA on- and off-target activities. As an increasing number of sequence features and characteristics are discovered and are incorporated into the MDL models, the prediction outcome is getting closer to experimental observations. We also introduced the basic principle of CRISPR activity and specificity and summarized the challenges they faced, aiming to facilitate the CRISPR communities to develop more accurate models for applying.

Motivation: Circular RNAs (circRNAs) are a new class of endogenous RNAs in animals and plants. During pre-RNA splicing, the 5' and 3' termini of exon(s) can be covalently ligated to form circRNAs through back-splicing (head-to-tail splicing). CircRNAs can be conserved across species, show tissue- and developmental stage-specific expression patterns, and may be associated with human disease. However, the mechanism of circRNA formation is still unclear although some sequence features have been shown to affect back-splicing., Results: In this study, by applying the state-of-art machine learning techniques, we have developed the first deep learning model, DeepCirCode, to predict back-splicing for human circRNA formation. DeepCirCode utilizes a convolutional neural network (CNN) with nucleotide sequence as the input, and shows superior performance over conventional machine learning algorithms such as support vector machine and random forest. Relevant features learnt by DeepCirCode are represented as sequence motifs, some of which match human known motifs involved in RNA splicing, transcription or translation. Analysis of these motifs shows that their distribution in RNA sequences can be important for back-splicing. Moreover, some of the human motifs appear to be conserved in mouse and fruit fly. The findings provide new insight into the back-splicing code for circRNA formation., Availability and Implementation: All the datasets and source code for model construction are available at https://github.com/BioDataLearning/DeepCirCode., Supplementary Information: Supplementary data are available at Bioinformatics online., (© The Author(s) 2019. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.)

High-grade gliomas are the most aggressive malignant brain tumors. Accurate pre-operative prognosis for this cohort can lead to better treatment planning. Conventional survival prediction based on clinical information is subjective and could be inaccurate. Recent radiomics studies have shown better prognosis by using carefully-engineered image features from magnetic resonance images (MRI). However, feature engineering is usually time consuming, laborious and subjective. Most importantly, the engineered features cannot effectively encode other predictive but implicit information provided by multi-modal neuroimages. We propose a two-stage learning-based method to predict the overall survival (OS) time of high-grade gliomas patient. At the first stage, we adopt deep learning, a recently dominant technique of artificial intelligence, to automatically extract implicit and high-level features from multi-modal, multi-channel preoperative MRI such that the features are competent of predicting survival time. Specifically, we utilize not only contrast-enhanced T1 MRI, but also diffusion tensor imaging (DTI) and resting-state functional MRI (rs-fMRI), for computing multiple metric maps (including various diffusivity metric maps derived from DTI, and also the frequency-specific brain fluctuation amplitude maps and local functional connectivity anisotropy-related metric maps derived from rs-fMRI) from 68 high-grade glioma patients with different survival time. We propose a multi-channel architecture of 3D convolutional neural networks (CNNs) for deep learning upon those metric maps, from which high-level predictive features are extracted for each individual patch of these maps. At the second stage, those deeply learned features along with the pivotal limited demographic and tumor-related features (such as age, tumor size and histological type) are fed into a support vector machine (SVM) to generate the final prediction result (i.e., long or short overall survival time). The experimental results demonstrate that this multi-model, multi-channel deep survival prediction framework achieves an accuracy of 90.66%, outperforming all the competing methods. This study indicates highly demanded effectiveness on prognosis of deep learning technique in neuro-oncological applications for better individualized treatment planning towards precision medicine.

A YOLOv5-based YOLOv5-KE unmanned aerial vehicle (UAV) image detection algorithm is proposed to address the low detection accuracy caused by the small size, high density, and overlapping leaves of Ambrosia trifida targets in UAV images. The YOLOv5-KE algorithm builds upon the YOLOv5 algorithm by adding a micro-scale detection layer, adjusting the hierarchical detection settings based on k-Means for Anchor Box, improving the loss function of CIoU, reselecting and improving the detection box fusion algorithm. Comparative validation experiments of the YOLOv5-KE algorithm for Ambrosia trifida recognition were conducted using a self-built dataset. The experimental results show that the best detection accuracy of Ambrosia trifida in UAV images is 93.9%, which is 15.2% higher than the original YOLOv5. Furthermore, this algorithm also outperforms other existing object detection algorithms such as YOLOv7, DC-YOLOv8, YOLO-NAS, RT-DETR, Faster RCNN, SSD, and Retina Net. Therefore, YOLOv5-KE is a practical algorithm for detecting Ambrosia trifida under complex field conditions. This algorithm shows good potential in detecting weeds of small, high-density, and overlapping leafy targets in UAV images, it could provide technical reference for the detection of similar plants.

Optical metasurfaces with pronounced spectral characteristics are promising for sensor applications. Currently, deep learning (DL) offers a rapid manner to design various metasurfaces. However, conventional DL models are usually assumed as black boxes, which is difficult to explain how a DL model learns physical features, and they usually predict optical responses of metasurfaces in a fuzzy way. This makes them incapable of capturing critical spectral features precisely, such as high quality (Q) resonances, and hinders their use in designing metasurface sensors. Here, a transformer‐based explainable DL model named Metaformer for the high‐intelligence design, which adopts a spectrum‐splitting scheme to elevate 99% prediction accuracy through reducing 99% training parameters, is established. Based on the Metaformer, all‐dielectric metasurfaces based on quasi‐bound states in the continuum (Q‐BIC) for high‐performance metasensing are designed, and fabrication experiments are guided potently. The explainable learning relies on spectral position encoding and multi‐head attention of meta‐optics features, which overwhelms traditional black‐box models dramatically. The meta‐attention mechanism provides deep physics insights on metasurface sensors, and will inspire more powerful DL design applications on other optical devices. [ABSTRACT FROM AUTHOR]

Predicting the dissolved oxygen concentration and identifying its driving factors are essential for improved prevention and management of anoxia in estuaries. However, complex hydrodynamic conditions and the limitations in traditional methods result in challenges in the identification of the driving factors for the low dissolved oxygen (DO) phenomenon. The objective of our study is to develop a robust deep learning model using four-year in situ data collected from an automatic water quality monitoring station (AWQMS) in an estuary, for accurate identification and quantification of the driving factors influencing DO levels. Mitigations in hypoxia were observed during the initial two years, but a subsequent decline in DO concentrations was witnessed recently. The periodicity of DO concentrations in the Pearl River Estuary reduced with the increase in the hypoxic intensity. Maximal information coefficient (MIC) and extreme gradient boosting (XGBoost) were employed to determine the significance of input variables, which were subsequently validated by using the long- and short-term memory networks (LSTMs). The driving factors contributing to the hypoxia problem were shown as temperature, pH, conductivity, and NH4+-N concentrations. Notably, the evaluation index values of the hybrid model are MAPE = 0.0887 and R2 = 0.9208, which have been improved compared with the LSTM model by about 99.34% in MAPE reduction and 16.56% in R2 improvement, indicating that the MixUp-LSTM model was capable of effectively capturing nonlinear relationships between DO and other water quality indicators. Based on existing literature, three traditional statistical methods and four machine learning models were also performed to compare with the proposed MixUp-LSTM model, which outperformed other models in terms of prediction accuracy and robustness. Overall, the successful identification of the driving factors for the deoxygenation phenomenon would have important implications for the governance and regulation of low DO in estuaries. [ABSTRACT FROM AUTHOR]