Front Big Data. 2025 Feb 20;8:1515341. doi: 10.3389/fdata.2025.1515341. eCollection 2025.

ABSTRACT

Neurodegenerative diseases are chronic, progressive conditions that cause irreversible damage to the nervous system, particularly in aging populations. Early diagnosis is a critical challenge, as these diseases often develop slowly and without clear symptoms until significant damage has occurred. Recent advances in radiomics and genomics have provided valuable insights into the mechanisms of these diseases by identifying specific imaging features and genomic patterns. Radiogenomics enhances diagnostic capabilities by linking genomics with imaging phenotypes, offering a more comprehensive understanding of disease progression. The growing field of artificial intelligence (AI), including machine learning and deep learning, opens new opportunities for improving the accuracy and timeliness of these diagnoses. This review examines the application of AI-based radiogenomics in neurodegenerative diseases, summarizing key model designs, performance metrics, publicly available data resources, significant findings, and future research directions. It provides a starting point and guidance for those seeking to explore this emerging area of study.

PMID:40052173 | PMC:PMC11882605 | DOI:10.3389/fdata.2025.1515341

Front Oncol. 2025 Feb 20;15:1536039. doi: 10.3389/fonc.2025.1536039. eCollection 2025.

ABSTRACT

With the rapid development of the "Internet + Medical" model, artificial intelligence technology has been widely used in the analysis of medical images. Among them, the technology of using deep learning algorithms to identify features of ultrasound and pathological images and realize intelligent diagnosis of diseases has entered the clinical verification stage. This study is based on the application research of artificial intelligence technology in medical diagnosis and reviews the early screening and diagnosis of thyroid diseases. The cure rate of thyroid disease is high in the early stage, but once it deteriorates into thyroid cancer, the risk of death and treatment costs of the patient increase. At present, the early diagnosis of the disease still depends on the examination equipment and the clinical experience of doctors, and there is a certain misdiagnosis rate. Based on the above background, it is particularly important to explore a technology that can achieve objective screening of thyroid lesions in the early stages. This paper provides a comprehensive review of recent research on the early diagnosis of thyroid diseases using artificial intelligence technology. It integrates the findings of multiple studies and that traditional machine learning algorithms are widely used as research objects. The convolutional neural network model has a high recognition accuracy for thyroid nodules and thyroid pathological cell lesions. U-Net network model can significantly improve the recognition accuracy of thyroid nodule ultrasound images when used as a segmentation algorithm. This article focuses on reviewing the intelligent recognition technology of thyroid ultrasound images and pathological sections, hoping to provide researchers with research ideas and help clinicians achieve intelligent early screening of thyroid cancer.

PMID:40052126 | PMC:PMC11882420 | DOI:10.3389/fonc.2025.1536039

Front Sports Act Living. 2025 Feb 20;7:1570591. doi: 10.3389/fspor.2025.1570591. eCollection 2025.

ABSTRACT

[This corrects the article DOI: 10.3389/fspor.2024.1512010.].

PMID:40051920 | PMC:PMC11882509 | DOI:10.3389/fspor.2025.1570591

Int J Gen Med. 2025 Mar 1;18:1173-1180. doi: 10.2147/IJGM.S507100. eCollection 2025.

ABSTRACT

PURPOSE: This study aims to evaluate the performance of a deep learning-based artificial intelligence (AI) diagnostic system in the analysis of retinal diseases, assessing its consistency with expert diagnoses and its overall utility in screening applications.

METHODS: A total of 3076 patients attending our hospital underwent comprehensive ophthalmic examinations. Initial assessments were performed using the AI, the Comprehensive AI Retinal Expert (CARE) system, followed by thorough manual reviews to establish final diagnoses. A comparative analysis was conducted between the AI-generated results and the evaluations by senior ophthalmologists to assess the diagnostic reliability and feasibility of the AI system in the context of ophthalmic screening.

RESULTS: : The AI diagnostic system demonstrated a sensitivity of 94.12% and specificity of 98.60% for diabetic retinopathy (DR); 89.50% sensitivity and 98.33% specificity for age-related macular degeneration (AMD); 91.55% sensitivity and 97.40% specificity for suspected glaucoma; 90.77% sensitivity and 99.10% specificity for pathological myopia; 81.58% sensitivity and 99.49% specificity for retinal vein occlusion (RVO); 88.64% sensitivity and 99.18% specificity for retinal detachment; 83.33% sensitivity and 99.80% specificity for macular hole; 82.26% sensitivity and 99.23% specificity for epiretinal membrane; 94.55% sensitivity and 97.82% specificity for hypertensive retinopathy; 83.33% sensitivity and 99.74% specificity for myelinated fibers; and 75.00% sensitivity and 99.95% specificity for retinitis pigmentosa. Additionally, the system exhibited notable performance in screening for other prevalent conditions, including DR, suspected glaucoma, suspected glaucoma, pathological myopia, and hypertensive retinopathy.

CONCLUSIONS: : The AI-assisted screening system exhibits high sensitivity and specificity for a majority of retinal diseases, suggesting its potential as a valuable tool for screening practices. Its implementation is particularly beneficial for grassroots and community healthcare settings, facilitating initial diagnostic efforts and enhancing the efficacy of tiered ophthalmic care, with important implications for broader clinical adoption.

PMID:40051895 | PMC:PMC11882464 | DOI:10.2147/IJGM.S507100

Eur Heart J Imaging Methods Pract. 2025 Mar 6;2(4):qyaf011. doi: 10.1093/ehjimp/qyaf011. eCollection 2024 Oct.

ABSTRACT

AIMS: Assessment of cardiac function is essential for diagnosis and treatment planning in cardiovascular disease. Volume of cardiac regions and the derived measures of stroke volume (SV) and ejection fraction (EF) are most accurately calculated from imaging. This study aims to develop a fully automatic deep learning approach for calculation of cardiac function from computed tomography (CT).

METHODS AND RESULTS: Time-resolved CT data sets from 39 patients were used to train segmentation models for the left side of the heart including the left ventricle (LV), left atrium (LA), and left atrial appendage (LAA). We compared nnU-Net, 3D TransUNet, and UNETR. Dice Similarity Scores (DSS) were similar between nnU-Net (average DSS = 0.91) and 3D TransUNet (DSS = 0.89) while UNETR performed less well (DSS = 0.69). Intra-class correlation analysis showed nnU-Net and 3D TransUNet both accurately estimated LVSV (ICCnnU-Net = 0.95; ICC3DTransUNet = 0.94), LVEF (ICCnnU-Net = 1.00; ICC3DTransUNet = 1.00), LASV (ICCnnU-Net = 0.91; ICC3DTransUNet = 0.80), LAEF (ICCnnU-Net = 0.95; ICC3DTransUNet = 0.81), and LAASV (ICCnnU-Net = 0.79; ICC3DTransUNet = 0.81). Only nnU-Net significantly predicted LAAEF (ICCnnU-Net = 0.68). UNETR was not able to accurately estimate cardiac function. Time to convergence during training and time needed for inference were both faster for 3D TransUNet than for nnU-Net.

CONCLUSION: nnU-Net outperformed two different vision transformer architectures for the segmentation and calculation of function parameters for the LV, LA, and LAA. Fully automatic calculation of cardiac function parameters from CT using deep learning is fast and reliable.

PMID:40051867 | PMC:PMC11883084 | DOI:10.1093/ehjimp/qyaf011

Chronic Dis Transl Med. 2024 Jun 9;11(1):1-21. doi: 10.1002/cdt3.137. eCollection 2025 Mar.

ABSTRACT

Chronic diseases such as heart disease, cancer, and diabetes are leading drivers of mortality worldwide, underscoring the need for improved efforts around early detection and prediction. The pathophysiology and management of chronic diseases have benefitted from emerging fields in molecular biology like genomics, transcriptomics, proteomics, glycomics, and lipidomics. The complex biomarker and mechanistic data from these "omics" studies present analytical and interpretive challenges, especially for traditional statistical methods. Machine learning (ML) techniques offer considerable promise in unlocking new pathways for data-driven chronic disease risk assessment and prognosis. This review provides a comprehensive overview of state-of-the-art applications of ML algorithms for chronic disease detection and prediction across datasets, including medical imaging, genomics, wearables, and electronic health records. Specifically, we review and synthesize key studies leveraging major ML approaches ranging from traditional techniques such as logistic regression and random forests to modern deep learning neural network architectures. We consolidate existing literature to date around ML for chronic disease prediction to synthesize major trends and trajectories that may inform both future research and clinical translation efforts in this growing field. While highlighting the critical innovations and successes emerging in this space, we identify the key challenges and limitations that remain to be addressed. Finally, we discuss pathways forward toward scalable, equitable, and clinically implementable ML solutions for transforming chronic disease screening and prevention.

PMID:40051825 | PMC:PMC11880127 | DOI:10.1002/cdt3.137

Front Psychiatry. 2025 Feb 20;16:1532256. doi: 10.3389/fpsyt.2025.1532256. eCollection 2025.

ABSTRACT

INTRODUCTION: Long-time mobile phone use (LTMPU) has been linked to emotional issues such as anxiety and depression while the enlarged perivascular spaces (EPVS), as marker of neuroinflammation, is closely related with mental disorders. In the current study, we aim to develop a predictive model utilizing MRI-quantified EPVS metrics and machine learning algorithms to assess the severity of anxiety and depression symptoms in patients with LTMPU.

METHODS: Eighty-two participants with LTMPU were included, with 37 suffering from anxiety and 44 suffering from depression. Deep learning algorithms were used to segment EPVS lesions and extract quantitative metrics. Comparison and correlation analyses were performed to investigate the relationship between EPVS and self-reported mood states. Training and testing datasets were randomly assigned in the ratio of 8:2 to perform radiomics analysis, where EPVS metrics combined with sex and age were used to select the most valuable features to construct machine learning models for predicting the severity of anxiety and depression.

RESULTS: Several EPVS features were significantly different between the two comparisons. For classifying anxiety status, eight features were selected to construct a logistic regression model, with an AUC of 0.819 (95%CI 0.573-1.000) in the testing dataset. For classifying depression status, eight features were selected to construct a K nearest neighbors model with an AUC value of 0.931 (95%CI 0.814-1.000) in the testing dataset.

DISCUSSION: The utilization of MRI-quantified EPVS metrics combined with machine-learning algorithms presents a promising method for evaluating severity of anxiety and depression symptoms in patients with LTMPU, which might introduce a non-invasive, objective, and quantitative approach to enhance diagnostic efficiency and guide personalized treatment strategies.

PMID:40051766 | PMC:PMC11882520 | DOI:10.3389/fpsyt.2025.1532256

Front Med (Lausanne). 2025 Feb 20;12:1497651. doi: 10.3389/fmed.2025.1497651. eCollection 2025.

ABSTRACT

BACKGROUND: The prognostic prediction of patients with hypercapnic respiratory failure holds significant clinical value. The objective of this study was to develop and validate a predictive model for predicting survival in patients with hypercapnic respiratory failure.

METHODS: The study enrolled a total of 697 patients with hypercapnic respiratory failure, including 565 patients from the First People's Hospital of Yancheng in the modeling group and 132 patients from the People's Hospital of Jiangsu Province in the external validation group. The three selected models were random survival forest (RSF), DeepSurv, a deep learning-based survival prediction algorithm, and Cox Proportional Risk (CoxPH). The model's predictive performance was evaluated using the C-index and Brier score. Receiver operating characteristic curve (ROC), area under ROC curve (AUC), and decision curve analysis (DCA) were employed to assess the accuracy of predicting the prognosis for survival at 6, 12, 18, and 24 months.

RESULTS: The RSF model (c-index: 0.792) demonstrated superior predictive ability for the prognosis of patients with hypercapnic respiratory failure compared to both the traditional CoxPH model (c-index: 0.699) and DeepSurv model (c-index: 0.618), which was further validated on external datasets. The Brier Score of the RSF model demonstrated superior performance, consistently measuring below 0.25 at the 6-month, 12-month, 18-month, and 24-month intervals. The ROC curve confirmed the superior discrimination of the RSF model, while DCA demonstrated its optimal clinical net benefit in both the modeling group and the external validation group.

CONCLUSION: The RSF model offered distinct advantages over the CoxPH and DeepSurv models in terms of clinical evaluation and monitoring of patients with hypercapnic respiratory failure.

PMID:40051730 | PMC:PMC11882423 | DOI:10.3389/fmed.2025.1497651

BMC Med Inform Decis Mak. 2025 Mar 6;25(1):113. doi: 10.1186/s12911-025-02924-w.

ABSTRACT

BACKGROUND: Dementia is a neurological syndrome marked by cognitive decline. Alzheimer's disease (AD) and frontotemporal dementia (FTD) are the common forms of dementia, each with distinct progression patterns. Early and accurate diagnosis of dementia cases (AD and FTD) is crucial for effective medical care, as both conditions have similar early-symptoms. EEG, a non-invasive tool for recording brain activity, has shown potential in distinguishing AD from FTD and mild cognitive impairment (MCI).

METHODS: This study aims to develop a deep learning-based classification system for dementia by analyzing EEG derived scout time-series signals from deep brain regions, specifically the hippocampus, amygdala, and thalamus. Scout time series extracted via the standardized low-resolution brain electromagnetic tomography (sLORETA) technique are utilized. The time series is converted to image representations using continuous wavelet transform (CWT) and fed as input to deep learning models. Two high-density EEG datasets are utilized to validate the efficacy of the proposed method: the online BrainLat dataset (128 channels, comprising 16 AD, 13 FTD, and 19 healthy controls (HC)) and the in-house IITD-AIIA dataset (64 channels, including subjects with 10 AD, 9 MCI, and 8 HC). Different classification strategies and classifier combinations have been utilized for the accurate mapping of classes in both data sets.

RESULTS: The best results were achieved using a product of probabilities from classifiers for left and right subcortical regions in conjunction with the DenseNet model architecture. It yield accuracies of 94.17 % and 77.72 % on the BrainLat and IITD-AIIA datasets, respectively.

CONCLUSIONS: The results highlight that the image representation-based deep learning approach has the potential to differentiate various stages of dementia. It pave the way for more accurate and early diagnosis, which is crucial for the effective treatment and management of debilitating conditions.

PMID:40050853 | DOI:10.1186/s12911-025-02924-w

BMC Med Imaging. 2025 Mar 6;25(1):77. doi: 10.1186/s12880-025-01616-1.

ABSTRACT

Automated segmentation of retinal blood vessels in fundus images plays a key role in providing ophthalmologists with critical insights for the non-invasive diagnosis of common eye diseases. Early and precise detection of these conditions is essential for preserving vision, making vessel segmentation crucial for identifying vascular diseases that pose a threat to vision. However, accurately segmenting blood vessels in fundus images is challenging due to factors such as significant variability in vessel scale and appearance, occlusions, complex backgrounds, variations in image quality, and the intricate branching patterns of retinal vessels. To overcome these challenges, the Unified Gated Swin Transformer with Multi-Feature Full Fusion (UGS-M3F) model has been developed as a powerful deep learning framework tailored for retinal vessel segmentation. UGS-M3F leverages its Unified Multi-Context Feature Fusion (UM2F) and Gated Boundary-Aware Swin Transformer (GBS-T) modules to capture contextual information across different levels. The UM2F module enhances the extraction of detailed vessel features, while the GBS-T module emphasizes small vessel detection and ensures extensive coverage of large vessels. Extensive experimental results on publicly available datasets, including FIVES, DRIVE, STARE, and CHAS_DB1, show that UGS-M3F significantly outperforms existing state-of-the-art methods. Specifically, UGS-M3F achieves a Dice Coefficient (DC) improvement of 2.12% on FIVES, 1.94% on DRIVE, 2.52% on STARE, and 2.14% on CHAS_DB1 compared to the best-performing baseline. This improvement in segmentation accuracy has the potential to revolutionize diagnostic techniques, allowing for more precise disease identification and management across a range of ocular conditions.

PMID:40050753 | DOI:10.1186/s12880-025-01616-1

BMC Med Imaging. 2025 Mar 6;25(1):78. doi: 10.1186/s12880-025-01610-7.

ABSTRACT

Deep learning methods have been migrated to rectal cancer staging as a classification process based on magnetic resonance images (MRIs). Typical approaches suffer from the imperceptible variation of images from different stage. The data augmentation also introduces scale invariance and rotation consistency problems after converting MRIs to 2D visible images. Moreover, the correctly labeled images are inadequate since T-staging requires pathological examination for confirmation. It is difficult for classification model to characterize the distinguishable features with limited labeled data. In this article, Laplace of Gaussian (LoG) filter is used to enhance the texture details of converted MRIs and we propose a new method named LoG-staging to predict the T stages of rectal cancer patients. We first use the LoG operator to clarify the fuzzy boundaries of rectal cancer cell proliferation. Then, we propose a new feature clustering method by leveraging the maximization of mutual information (MMI) mechanism which jointly learns the parameters of a neural network and the cluster assignments of features. The assignments are used as labels for the next round of training, which compensate the inadequacy of labeled training data. Finally, we experimentally verify that the LoG-staging is more accurate than the nonlinear dimensionality reduction in predicting the T stages of rectal cancer. We innovatively implement information bottleneck (IB) method in T-staging of rectal cancer based on image classification and impressive results are obtained.

PMID:40050741 | DOI:10.1186/s12880-025-01610-7

NPJ Digit Med. 2025 Mar 6;8(1):144. doi: 10.1038/s41746-025-01483-8.

ABSTRACT

The DESTINY-Breast04 trial has recently demonstrated survival benefits of trastuzumab-deruxtecan (T-DXd) in metastatic breast cancer patients with low Human Epidermal Growth Factor Receptor 2 (HER2) expression. Accurate differentiation of HER2 scores has now become crucial. However, visual immunohistochemistry (IHC) scoring is labour-intensive and prone to high interobserver variability, and artificial intelligence (AI) has emerged as a promising tool in diagnostic medicine. We conducted a diagnostic meta-analysis to evaluate AI's performance in classifying HER2 IHC scores, demonstrating high accuracy in predicting T-DXd eligibility, with a pooled sensitivity of 0.97 [95% CI 0.96-0.98] and specificity of 0.82 [95% CI 0.73-0.88]. Meta-regression revealed better performance with deep learning and patch-based analysis, while performance declined in externally validated and those utilising commercially available algorithms. Our findings indicate that AI holds promising potential in accurately identifying HER2-low patients and excels in distinguishing 2+ and 3+ scores.

PMID:40050686 | DOI:10.1038/s41746-025-01483-8

Sci Rep. 2025 Mar 6;15(1):7817. doi: 10.1038/s41598-025-92582-9.

ABSTRACT

This study presents a novel hybrid deep learning model for arrhythmia classification from electrocardiogram signals, utilizing the stockwell transform for feature extraction. As ECG signals are time-series data, they are transformed into the frequency domain to extract relevant features. Subsequently, a CNN is employed to capture local patterns, while a transformer architecture learns long-term dependencies. Unlike traditional CNN-based models that require R-peak detection, the proposed model operates without it and demonstrates superior accuracy and efficiency. The findings contribute to enhancing the accuracy of ECG-based arrhythmia diagnosis and are applicable to real-time monitoring systems. Specifically, the model achieves an accuracy of 97.8% on the Icentia11k dataset using four arrhythmia classes and 99.58% on the MIT-BIH dataset using five arrhythmia classes.

PMID:40050678 | DOI:10.1038/s41598-025-92582-9

Sci Rep. 2025 Mar 6;15(1):7876. doi: 10.1038/s41598-025-92594-5.

ABSTRACT

Whole-body bone scan (WBS) is usually used as the effective diagnostic method for early-stage and comprehensive bone metastases of breast cancer. WBS images with breast cancer bone metastasis have the characteristics of low resolution, small foreground, and multiple lesions, hindering the widespread application of deep learning-based models. Automatically detecting a large number of densely small lesions on low-resolution WBS images remains a challenge. We aim to develop a unified framework for detecting multiple densely bone metastases based on low-resolution WBS images. We propose a novel unified detection framework to detect multiple bone metastases based on WBS images. Considering the difficulties of feature extraction caused by low resolution and multiple lesions, we innovatively propose the plug-and-play position auxiliary extraction module and feature fusion module to enhance the ability of global information extraction. In order to accurately detect small metastases in WBS, we designed the self-attention transformer-based target detection head. This retrospective study included 512 patients with breast cancer bone metastases from Peking Union Medical College Hospital. The data type is whole-body bone scan image. For our study, the ratio of training set, validation set and test set is about 6:2:2. The benchmarks are four representative baselines, SSD, YOLOR, Faster_RCNN_R and Scaled-YOLOv4. The performance metrics are Average Precision (AP), Precision and Recall. The detection results obtained through the proposed method were assessed using the Bonferroni-adjusted Wilcoxon rank test. The significant level is adjusted according to different multiple comparisons. We conducted extensive experiments and ablation studies on a private dataset of breast cancer WBS and a public dataset of bone scans from West China Hospital to validate the effectiveness and generalization. Experiments were conducted to evaluate the effectiveness of our method. First, compared to different network architectures, our method obtained AP of 55.0 ± 6.4% (95% confidence intervals (CI) 49.9-60.1%, [Formula: see text]), which improved AP by 45.2% for the SSD baseline with AP 9.8 ± 2% (95% CI 8.1-11.4%). For the metric of recall, our method achieved the average of 54.3 ± 4.2% (95% CI 50.9-57.6%, [Formula: see text]), which has improved the recall values by 49.01% for the SSD model with 5.2 ± 12.7% (95% CI 10-21.3%). Second, we conducted ablation studies. On the private dataset, adding the detection head module and position auxiliary extraction module will increase the AP values by 14.03% (from 33.3 ± 2% to 47.6 ± 4.4%) and 19.3% (from 33.3 ± 2% to 52.6 ± 6.1%), respectively. In addition, the generalization of the method was also verified on the public dataset BS-80K from West China Hospital. Extensive experimental results have demonstrated the superiority and effectiveness of our method. To the best of our knowledge, our work is the first attempt for developing automatic detector considering the unique characteristics of low resolution, small foreground and multiple lesions of breast cancer WBS images. Our framework is tailored for whole-body WBS and can be used as a clinical decision support tool for early decision-making for breast cancer bone metastases.

PMID:40050676 | DOI:10.1038/s41598-025-92594-5

Nat Commun. 2025 Mar 6;16(1):2269. doi: 10.1038/s41467-025-57485-3.

ABSTRACT

Pharmacophores are abstractions of essential chemical interaction patterns, holding an irreplaceable position in drug discovery. Despite the availability of many pharmacophore tools, the adoption of deep learning for pharmacophore-guided drug discovery remains relatively rare. We herein propose a knowledge-guided diffusion framework for 'on-the-fly' 3D ligand-pharmacophore mapping, named DiffPhore. It leverages ligand-pharmacophore matching knowledge to guide ligand conformation generation, meanwhile utilizing calibrated sampling to mitigate the exposure bias of the iterative conformation search process. By training on two self-established datasets of 3D ligand-pharmacophore pairs, DiffPhore achieves state-of-the-art performance in predicting ligand binding conformations, surpassing traditional pharmacophore tools and several advanced docking methods. It also manifests superior virtual screening power for lead discovery and target fishing. Using DiffPhore, we successfully identify structurally distinct inhibitors for human glutaminyl cyclases, and their binding modes are further validated through co-crystallographic analysis. We believe this work will advance the AI-enabled pharmacophore-guided drug discovery techniques.

PMID:40050649 | DOI:10.1038/s41467-025-57485-3

Sci Rep. 2025 Mar 6;15(1):7900. doi: 10.1038/s41598-025-90415-3.

ABSTRACT

Cervical cancer (CC) is the leading cancer, which mainly affects women worldwide. It generally occurs from abnormal cell evolution in the cervix and a vital functional structure in the uterus. The importance of timely recognition cannot be overstated, which has led to various screening methods such as colposcopy, Human Papillomavirus (HPV) testing, and Pap smears to identify potential threats and enable early intervention. Early detection during the precancerous phase is crucial, as it provides an opportunity for effective treatment. The diagnosis and screening of CC depend on colposcopy and cytology models. Deep learning (DL) is an appropriate technique in computer vision, which has developed as a latent solution to increase the efficiency and accuracy of CC screening when equated to conventional clinical inspection models that are vulnerable to human error. This study presents a Leveraging Swin Transformer with an Ensemble of Deep Learning Model for Cervical Cancer Screening (LSTEDL-CCS) technique for colposcopy images. The presented LSTEDL-CCS technique aims to detect and classify CC on colposcopy images. Initially, the wiener filtering (WF) model is used for image pre-processing. Next, the swin transformer (ST) network is utilized for feature extraction. For the cancer detection process, the ensemble learning method is performed by employing three models, namely autoencoder (AE), bidirectional gated recurrent unit (BiGRU), and deep belief network (DBN). Finally, the hyperparameter tuning of the DL techniques is performed using the Pelican Optimization Algorithm (POA). A comprehensive experimental analysis is conducted, and the results are evaluated under diverse metrics. The performance validation of the LSTEDL-CCS methodology portrayed a superior accuracy value of 99.44% over existing models.

PMID:40050635 | DOI:10.1038/s41598-025-90415-3

Sci Data. 2025 Mar 6;12(1):390. doi: 10.1038/s41597-025-04701-w.

ABSTRACT

Delineating the extent of urban villages (UVs) is crucial for effective urban planning and management, as well as for providing targeted policy and financial support. Unlike field surveys, the interpretation of satellite imagery provides an efficient, near real-time, and objective means of mapping UV. However, current research efforts predominantly concentrate on individual cities, resulting in a scarcity of interpretable UV maps for numerous other cities. This gap in availability not only hinders public awareness of the distribution and evolution of UV but also limits the reliability and transferability of models due to the insufficient number and diversity of samples. To address this issue, we developed CUGUV, a benchmark dataset that includes a diverse collection of thousands of UV samples, carefully curated from 15 major cities across various geographical regions in China. The dataset can be accessed through this link: https://doi.org/10.6084/m9.figshare.26198093 . This dataset can serve as a foundation for evaluating and improving the robustness and transferability of models. Subsequently, we present an innovative framework that effectively integrates and learns from multiple data sources to better address the cross-city UV mapping task. Tests show that the proposed models achieve over 92% in overall accuracy, precision, and F1-scores, outperforming state-of-the-art models. This highlights the effectiveness of both the proposed dataset and model. This presented dataset and model bolsters our capability to better understand and accurately model these complex and diverse phenomena, ultimately leading to a notable improvement in the performance of large-scale UV mapping.

PMID:40050634 | DOI:10.1038/s41597-025-04701-w

Nat Commun. 2025 Mar 6;16(1):2251. doi: 10.1038/s41467-025-57516-z.

ABSTRACT

Frequency transfer is a key challenge in machine learning as it allows researchers to go beyond in-range analyses of spectrum properties towards out-of-the-range predictions. Traditionally, to predict properties at a specific frequency, targeted spectrum is included in training data for a deep neural network (DNN). However, due to limitations of measurement or computation source, training data at some frequencies are hardly accessible, especially for multi-physics problems. In this work, we propose a multi-physics deep learning framework (MDLF) consisting of a multi-fidelity DeepONet, a Euler latent dynamic network, and a data-analytical inversion network. Without the knowledge about multi-physics response, MDLF is successfully generalized to unseen frequency bands for both parametric and free-form metasurface by dynamically utilizing a Euler latent space and single-physics information. Moreover, an inversion method is introduced to incorporate hybrid a priori in inverse design of metasurface. Under EM-thermal coupling, we verify the proposed MDLF numerically and experimentally.

PMID:40050630 | DOI:10.1038/s41467-025-57516-z

ACS Appl Mater Interfaces. 2025 Mar 6. doi: 10.1021/acsami.4c22848. Online ahead of print.

ABSTRACT

In the past decades, conjugated polyelectrolytes (CPEs) have become prominent in sensing applications due to their unique properties, including strong and tunable light absorption, high sensitivity, water solubility, and biocompatibility. Inspired by mammalian olfactory and gustatory systems, CPE-based sensor arrays have made significant strides in discriminating structurally similar analytes and complex mixtures for various applications. This review consolidates recent advancements in CPE-based sensor arrays, highlighting rational design, controllable fabrication, and effective data processing methods. It covers the fundamentals of CPE fluorescence sensing, emphasizing design strategies for sensor array units and data processing techniques. The broad applicability of CPE-based sensor arrays is demonstrated across diverse domains, including environmental monitoring (e.g., detecting metal ions and explosives), medical diagnostics (e.g., sensing disease markers and analyzing biological samples), and food safety (e.g., assessing the freshness, quality, and source of food products). Further, challenges and future directions in the field are discussed to inspire further research and development in this area.

PMID:40048404 | DOI:10.1021/acsami.4c22848

IEEE Trans Pattern Anal Mach Intell. 2025 Mar 6;PP. doi: 10.1109/TPAMI.2025.3548148. Online ahead of print.

ABSTRACT

Single Image Reflection Removal (SIRR) is a canonical blind source separation problem and refers to the issue of separating a reflection-contaminated image into a transmission and a reflection image. The core challenge lies in minimizing the commonalities among different sources. Existing deep learning approaches either neglect the significance of feature interactions or rely on heuristically designed architectures. In this paper, we propose a novel Deep Exclusion unfolding Network (DExNet), a lightweight, interpretable, and effective network architecture for SIRR. DExNet is principally constructed by unfolding and parameterizing a simple iterative Sparse and Auxiliary Feature Update (i-SAFU) algorithm, which is specifically designed to solve a new model-based SIRR optimization formulation incorporating a general exclusion prior. This general exclusion prior enables the unfolded SAFU module to inherently identify and penalize commonalities between the transmission and reflection features, ensuring more accurate separation. The principled design of DExNet not only enhances its interpretability but also significantly improves its performance. Comprehensive experiments on four benchmark datasets demonstrate that DExNet achieves state-of-the-art visual and quantitative results while utilizing only approximately 8% of the parameters required by leading methods.

PMID:40048344 | DOI:10.1109/TPAMI.2025.3548148