With the intensification of global population aging, the incidence of cognitive disorders such as dementia continues to rise. The Mini-Mental State Examination (MMSE) and other alternative tools can help doctors detect subtle changes in cognitive function at an early stage. These assessment tools can make a diagnosis before symptoms become severe, providing opportunities for early intervention, which is crucial for delaying disease progression and improving the quality of life of patients. However, traditional cognitive assessment methods are overly complex and affected by various factors. With the development of artificial intelligence technology, many new assessment tools are constantly being developed and improved. How to evaluate the effectiveness of intelligent electronic cognitive assessment tools is particularly important. We have proposed the Correlation and Supervised Learning-based Cognitive Tool Effectiveness Assessment Method (CSL-CTEA) to evaluate the effectiveness of intelligent electronic cognitive assessment tools, including: (1) experimental design and data collection based on traditional scales and intelligent electronic assessment tools, (2) consistency and correlation tests; (3) accuracy analysis of assessment results based on supervised learning. We used CSL-CTEA to explore the effectiveness of a certain electronic assessment. This intelligent electronic cognitive assessment tool includes voice tests, orientation tests, and picture recognition tests to assess cognitive abilities from multiple perspectives. The results show that the electronic assessment is in good agreement with traditional cognitive assessment methods. The various indicators of the electronic assessment can explain the changes in MMSE scores to some extent. The study also found that the electronic assessment performs well in determining whether the subject is at cognitive risk. To some extent, the electronic assessment can replace traditional cognitive assessment methods such as MMSE to help people judge whether they are at risk of cognitive decline.

The study aimed to diagnose of Alzheimer's Disease (AD) and Frontotemporal Dementia (FTD) based on brain functional connectivity features extracted via resting-state Electroencephalographic (EEG) signals, and subsequently developed a convolutional neural network (CNN) model, Coherence-CNN, for classification. First, a publicly available dataset of EEG resting state-closed eye recordings containing 36 AD subjects, 23 FTD subjects, and 29 cognitively normal (CN) subjects was used. Then, coherence metrics were utilized to quantify brain functional connectivity, and the differences in coherence between groups across various frequency bands were investigated. Next, spectral clustering was used to analyze variations and differences in brain functional connectivity related to disease states, revealing distinct connectivity patterns in brain electrode position maps. The results demonstrated that brain functional connectivity between different regions was more robust in the CN group, while the AD and FTD groups exhibited various degrees of connectivity decline, reflecting the pronounced differences in connectivity patterns associated with each condition. Furthermore, Coherence-CNN was developed based on CNN and the feature of coherence for three-class classification, achieving a commendable accuracy of 94.32% through leave-one-out cross-validation. This study revealed that Coherence-CNN demonstrated significant performance for distinguishing AD, FTD, and CN groups, supporting the disorder of brain functional connectivity in AD and FTD.

Multivariate time series forecasting (MTSF) aims to predict time series data containing multiple variates, which requires the consideration of both intra-series temporal trends and inter-series interactions. Benefiting from the success of Transformers in computer vision and natural language processing, recently, many time series models are based on the Transformer architecture to reveal intra-series or/and inter-series relationships. However, limited by a single time series embedding representation, most existing methods focus either on the study of changes in the time dimension or on the interactions between multiple variates. There is still a significant research gap in the comprehensive understanding of inter-series interactions and the changes within intra-series in multivariate time series. To bridge this gap, we propose a dual-stream interactive networks with pearson-mask awareness (DSIN-PMA) for MTSF. Specifically, we first employ a dual-stream embedding structure with multivariate embedding and time-step embedding to better represent the diversity in the time and the variate dimensions, respectively. Furthermore, the overall framework of the model adopts a two-stream networks: (1) A cross-multivariate attention with pearson-mask module is proposed to effectively mitigate the impact of noise in multivariate time series data and reduce unnecessary dispersion in cross-variate interactive attention, thereby helping to learn the dependencies between multiple variates more efficiently. (2) A time-step attention module is introduced to learn seasonality and potential trends in the time dimension. Finally, to further enhance the feature representation ability and robustness of the model, we employ a cross-dimension consistency learning strategy to interact with the outputs of the dual-stream encoder. Experimental results on 11 real-world data sets from multiple fields show that the DSIN-PMA model significantly outperforms the baseline model, and achieves 5.12%-17.43% improvements over state-of-the-art (SOTA) methods. In-depth analysis shows that the strategy that comprehensively considers both the variate dimension and the time-step dimension outperforms the single-dimensional strategy. Our source codes are publicly available at https://github.com/yejunjiePhD/DSIN-PMA.

Developing efficient and accurate quantitative analysis methods for plastic blends holds significant value for resource recycling and environmental monitoring. Mid-infrared (MIR) spectroscopy, combined with chemometric techniques, has demonstrated excellent performance in plastic blend quantification. However, obtaining mid-infrared spectral data for a large number of plastic blends to calibrate the model remains challenging. This study proposes an innovative approach that utilizes pure plastic MIR spectra and the Beer-Lambert law to generate virtual plastic blend spectra. Four experimental groups (A-D) were designed, incorporating both real and virtual spectra to systematically evaluate the method's effectiveness. Experiment group A established a baseline model using real spectra, while experiment groups B, C, and D respectively validated the applicability and generalization capability of models based on virtual spectra, as well as their potential applications in MIR hyperspectral imaging (MIR-HSI), respectively. The study further explores feature band selection, model construction, evaluation, and interpretation. The results demonstrate that this method can efficiently predict the mass percentages of components in ternary plastic blends. In experimental group C, partial least squares regression (PLSR), one-dimensional convolutional neural network (CNN1D), and two-dimensional convolutional neural network based on Gramian Angular Field (GAF-CNN2D) models-trained on 208 virtual plastic blend spectra-were employed to predict the mass percentages of 66 ternary plastic blends composed of polyethylene (PE), polypropylene (PP), and polystyrene (PS). The prediction coefficients of determination (RT2) reached 0.9872, 0.9879, and 0.9944, respectively, indicating exceptional predictive accuracy. Experimental group D further demonstrated that, even under Gaussian noise interference and limited spectral range, the strategy of fusing mid-wave infrared and long-wave infrared bands allowed the PLSR and GAF-CNN2D models to maintain high performance in predicting the mass percentages of 66 ternary blends of PE, PP, and PS, with RT2 values of 0.9852 and 0.9895, respectively. This suggests that the proposed method holds potential for applications in MIR-HSI and is promising for real-time online analysis. Finally, this study proposes a more widely applicable and optimized quantitative analysis application scheme based on virtual plastic blend spectra, aiming to enable rapid and precise determination of unknown plastic blends.

The vast volumes of data are needed to train Deep Learning Models from scratch to identify illnesses in soybean leaves. However, there is still a lack of sufficient high-quality samples. To overcome this problem, we have developed the real-life SoyLeaf dataset and used the pre-trained Deep Learning Models to identify leaf diseases. In this paper, we have initially developed the real-life SoyLeaf dataset collected from the ICAR-Indian Institute of Soybean Research (IISR) Center, Indore field. This SoyLeaf dataset contains 9786 high-quality soybean leaf images, including healthy and diseased leaves. Following this, we have adapted data preprocessing techniques to enhance the quality of images. In addition, we have utilized several Deep Learning Models, i.e., fourteen Keras Transfer Learning Models, to determine which model best fits the dataset on SoyLeaf diseases. The accuracies of the proposed fine-tuned models using the Adam optimizer are as follows: ResNet50V2 achieves 99.79%, ResNet101V2 achieves 99.89%, ResNet152V2 achieves 99.59%, InceptionV3 achieves 99.83%, InceptionResNetV2 achieves 99.79%, MobileNet achieves 99.82%, MobileNetV2 achieves 99.89%, DenseNet121 achieves 99.87%, and DenseNet169 achieves 99.87%. Similarly, the accuracies of the proposed fine-tuned models using the RMSprop optimizer are as follows: ResNet50V2 achieves 99.49%, ResNet101V2 achieves 99.45%, ResNet152V2 achieves 99.45%, InceptionV3 achieves 99.58%, InceptionResNetV2 achieves 99.88%, MobileNet achieves 99.73%, MobileNetV2 achieves 99.83%, DenseNet121 achieves 99.89%, and DenseNet169 achieves 99.77%. The experimental results of the proposed fine-tuned models show that only ResNet50V2, ResNet101V2, InceptionV3, InceptionResNetV2, MobileNet, MobileNetV2, DenseNet121, and DenseNet169 have performed better in terms of training, validation, and testing accuracies than other state-of-the-art models.

Lubrication plays a pivotal role in modern society, given its significant economic and environmental implications, particularly in relation to friction, wear and the failure of moving mechanical systems. With recent breakthroughs in computational architectures, the development of advanced simulation frameworks has been greatly accelerated, facilitating the study of surfaces, lubricants and additives at unprecedented scales. However, the inherently multiscale nature of lubricated contacts necessitates a delicate balance between computationally efficient continuum descriptions and detailed atomistic accuracy for addressing the complex physiochemical phenomena spanning vastly different spatiotemporal scales. This review explores the dilemma of modelling inherently multiphysics tribological interactions, which drive the evolution of lubricated interfaces and shape tribosystem performances across the scales as accurately and simultaneously as efficiently as possible. It critically examines state-of-the-art modelling tools, their applications and limitations across spatiotemporal domains. Moreover, the capacity for machine learning to aggregate extensive datasets, address multi-physical complexities ranging from atomic dimensions to macroscopic scales and accelerate simulation workflows is explored, offering transformative perspectives for the future of lubrication modelling.

In ophthalmology, artificial intelligence (AI) demonstrates potential in using ophthalmic imaging across diverse diseases, often matching ophthalmologists' performance. However, the range of machine learning models for epiretinal membrane (ERM) management, which differ in methodology, application, and performance, remains largely unsynthesized.

Epiretinal membrane management relies on clinical evaluation and imaging, with surgical intervention considered in cases of significant impairment. AI analysis of ophthalmic images and clinical features could enhance ERM detection, characterization, and prognostication, potentially improving clinical decision-making. This scoping review aims to evaluate the methodologies, applications, and reported performance of AI models in ERM diagnosis, characterization, and prognostication.

A comprehensive literature search was conducted across 5 electronic databases including Ovid MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, and Web of Science Core Collection from inception to November 14, 2024. Studies pertaining to AI algorithms in the context of ERM were included. The primary outcomes measured will be the reported design, application in ERM management, and performance of each AI model.

Three hundred ninety articles were retrieved, with 33 studies meeting inclusion criteria. There were 30 studies (91%) reporting their training and validation methods. Altogether, 61 distinct AI models were included. OCT scans and fundus photographs were used in 26 (79%) and 7 (21%) papers, respectively. Supervised learning and both supervised and unsupervised learning were used in 32 (97%) and 1 (3%) studies, respectively. Twenty-seven studies (82%) developed or adapted AI models using images, whereas 5 (15%) had models using both images and clinical features, and 1 (3%) used preoperative and postoperative clinical features without ophthalmic images. Study objectives were categorized into 3 stages of ERM care. Twenty-three studies (70%) implemented AI for diagnosis (stage 1), 1 (3%) identified ERM characteristics (stage 2), and 6 (18%) predicted vision impairment after diagnosis or postoperative vision outcomes (stage 3). No articles studied treatment planning. Three studies (9%) used AI in stages 1 and 2. Of the 16 studies comparing AI performance to human graders (i.e., retinal specialists, general ophthalmologists, and trainees), 10 (63%) reported equivalent or higher performance.

Artificial intelligence-driven assessments of ophthalmic images and clinical features demonstrated high performance in detecting ERM, identifying its morphological properties, and predicting visual outcomes following ERM surgery. Future research might consider the validation of algorithms for clinical applications in personal treatment plan development, ideally to identify patients who might benefit most from surgery.

The author(s) have no proprietary or commercial interest in any materials discussed in this article.

Seed viability, a key indicator for quality assessment, directly impacts the emergence of field seedlings. The existing nondestructive testing model for maize seed vitality based on naturally aged seeds and predominantly relying on single-modal data like MV and RS, achieves an accuracy of less than 70 %. To elucidate the influence of different data on model accuracy, this study proposes the MSCNSVN model for detecting seed viability by collecting multisensor information from maize seeds using sensors, such as MV, RS, TS, FS, and SS. Our findings indicated that (1) the single-modal FS dataset achieved optimal prediction accuracy, with FS570/600 contributing the most; (2) multimodal data fusion outperformed single-modal data, with an accuracy improvement of 10 %, while the MV + RS + FS dataset achieved the highest accuracy; (3) the MSCNSVN model demonstrated superior performance compared to baseline models; (4) modeling with dual-variety datasets and endosperm surface datasets improved accuracy by 2 %-3 %.

Pulmonary arterial hypertension (PAH) significantly affects the pulmonary vasculature, requiring accurate estimation of mean pulmonary arterial pressure (mPAP) and pulmonary vascular resistance index (PVRi). Although cardiac catheterization is the gold standard for these measurements, it poses risks, especially in children. This pilot study explored how machine learning (ML) can predict pulmonary hemodynamics from non-invasive cardiac magnetic resonance (CMR) cine images in pediatric PAH patients.

A retrospective analysis of 40 CMR studies from children with PAH using a four-fold stratified group cross-validation was conducted. The endpoints were severity profiles of mPAP and PVRi, categorised as 'low', 'high', and 'extreme'. Deep learning (DL) and traditional ML models were optimized through hyperparameter tuning. Receiver operating characteristic curves and area under the curve (AUC) were used as the primary evaluation metrics.

DL models utilizing CMR cine imaging showed the best potential for predicting mPAP and PVRi severity profiles on test folds (AUCmPAP=0.82 and AUCPVRi=0.73). True positive rates (TPR) for predicting low, high, and extreme mPAP were 5/10, 11/16, and 11/14, respectively. TPR for predicting low, high, and extreme PVRi were 5/13, 14/15, and 7/12, respectively. Optimal DL models only used spatial patterns from consecutive CMR cine frames to maximize prediction performance.

This exploratory pilot study demonstrates the potential of DL leveraging CMR imaging for non-invasive prediction of mPAP and PVRi in pediatric PAH. While preliminary, these findings may lay the groundwork for future advancements in CMR imaging in pediatric PAH, offering a pathway to safer disease monitoring and reduced reliance on invasive cardiac catheterization.

Large Language Models (LLMs), recognized for their advanced capabilities in natural language processing, have been successfully employed across various domains. However, their effectiveness in addressing challenges related to drug discovery has yet to be fully elucidated.

In this paper, we propose a novel LLM based method for drug-drug interaction (DDI) prediction, named DDI-JUDGE, achieved through the integration of judging and ICL prompts. The proposed method outperforms existing LLM approaches, demonstrating the potential of LLMs for predicting DDIs. We introduce a novel in-context learning (ICL) prompt paradigm that selects high-similarity samples as positive and negative prompts, enabling the model to effectively learn and generalize knowledge. Additionally, we present an ICL-based prompt template that structures inputs, prediction tasks, relevant factors, and examples, leveraging the pre-trained knowledge and contextual understanding of LLMs to enhance DDI prediction capabilities. To further refine predictions, we employ GPT-4 as a discriminator to assess the relevance of predictions generated by multiple LLMs.

DDI-JUDGE achieves the best performance among all models in both zero-shot and few-shot settings, with an AUC of 0.642/0.788 and AUPR of 0.629/0.801, respectively. These results demonstrate its superior predictive capability and robustness across different learning scenarios.

These findings highlight the potential of LLMs in advancing drug discovery through more effective DDI prediction. The modular prompt structure, combined with ensemble reasoning, offers a scalable framework for knowledge-intensive biomedical applications. The code for DDI-JUDGE is available at https://github.com/zcc1203/ddi-judge.

Femtosecond laser-induced grating scattering (fs-LIGS) technique has emerged as a robust diagnostic technique for combustion flow field characterization, particularly in temperature and pressure measurements. This study investigates four deep learning architectures for the simultaneous prediction of gas-phase temperature and pressure from raw fs-LIGS signals, including fully connected neural networks (FCNN), convolutional neural networks (CNN), bidirectional long short-term memory networks (BiLSTM), and a hybrid CNN-BiLSTM model. All architectures demonstrated exceptional predictive accuracy, achieving mean percentage errors ranging from -3.94% to 0.32% for temperature estimation and -0.26% to 1.09% for pressure estimation. The detailed comparison of the four models suggests that the LSTM structures have better adaptability for the measurements of temperature and pressure. In contrast the, CNN-BiLSTM model demonstrated superior overall performance in terms of predictive accuracy, cross-validation robustness, convergence rate, memory requirements, etc. Successful integration of these deep learning approaches with an fs-LIGS system would establish a novel framework for real-time, multi-parameter combustion diagnostics.

Bioprinting enables the creation of complex tissue scaffolds, which are vital for tissue engineering. However, predicting scaffold biocompatibility before fabrication remains a critical challenge, potentially leading to inefficiencies and resource wastage. Artificial Intelligence (AI) models, particularly Artificial Neural Networks (ANNs) and Convolutional Neural Networks (CNNs), offer promising predictive capabilities to address this issue. This study aims to compare the performance of ANN and CNN models to identify the most suitable approach for predicting scaffold biocompatibility using PrusaSlicer-generated designs.

Fifteen key design parameters influencing scaffold biocompatibility were modelled using ANN, while scaffold images were analyzed using CNN. PrusaSlicer was employed in designing scaffolds, with parameters influencing biocompatibility predictions. ANN models analyzed these parameters, while CNN models processed scaffold images. Data was standardized, and models were trained on an 80/20 split dataset. Performance evaluation metrics included accuracy, precision, recall, F1-Scores, and confusion matrices. Experimental validation involved biocompatibility tests on five scaffolds.

ANN model with 20 neurons and 100 epochs earned perfect (1.0) scores in F1-Score, Precision, and Recall, indicating the best possible model performance. A batch size of 56 for the Convolutional Neural Network model demonstrated balance in F1-Score (0.87), Precision (0.88), and Recall (0.9). Five scaffold tissues were tested for biocompatibility using these two models. ANN model predicted 5 scaffold tissues' biocompatibilities correctly. While the ANN model accurately predicted biocompatibilities for all five scaffold samples, the CNN model misclassified one sample.

This study demonstrates that ANN models are superior to CNN models in predicting scaffold biocompatibility from numerical design parameters. The findings underscore the value of ANNs for structured data in bioprinting, enhancing prediction accuracy and efficiency. These insights can accelerate advancements in tissue engineering and personalized medicine by reducing costs and improving success rates in bioprinting applications. Future work will focus on addressing overfitting challenges and optimizing the models to further enhance their robustness and predictive capabilities.

Acute lymphoblastic leukemia (ALL) is a leading cause of death among pediatric malignancies. Early diagnosis of ALL is crucial for minimizing misdiagnosis, improving survival rates, and ensuring the implementation of precise treatment plans for patients.

In this study, we propose a multi-modal deep neural network-based framework for early and efficient screening of ALL. Both white blood cell (WBC) scattergrams and complete blood count (CBC) are employed for ALL detection. The dataset comprises medical data from 233 patients with ALL, 283 patients with infectious mononucleosis (IM), and 183 healthy controls (HCs).

The combination of CBC data with WBC scattergrams achieved an accuracy of 98.43% in fivefold cross-validation and a sensitivity of 96.67% in external validation, demonstrating the efficacy of our method. Additionally, the area under the curve (AUC) of this model surpasses 0.99, outperforming well-trained medical technicians.

To the best of our knowledge, this framework is the first to incorporate WBC scattergrams with CBC data for ALL screening, proving to be an efficient method with enhanced sensitivity and specificity. Integrating this framework into the screening procedure shows promise for improving the early diagnosis of ALL and reducing the burden on medical technicians. The code and dataset are available at https://github.com/cvi-szu/ALL-Screening.

Cancer continues to pose significant challenges in healthcare due to the complex nature of tumor progression. In this digital era, artificial intelligence has emerged as a powerful tool that can potentially transform multiple aspects of cancer care.

In the current study, we conducted a comprehensive literature search across databases such as PubMed, Scopus, and IEEE Xplore. Studies published between 2014 and 2024 were considered. The selection process involved a systematic screening based on predefined inclusion and exclusion criteria. Studies were included if they focused on applying machine learning techniques for tumor progression modeling, diagnosis, or prognosis, were published in peer-reviewed journals or conference proceedings, were available in English, and presented experimental results, simulations, or real-world applications. In total, 87 papers were included in this review, ensuring a diverse and representative analysis of the field. A workflow is included to illustrate the procedure followed to achieve this aim.

This review delves into the cutting-edge applications of machine learning (ML), including supervised learning methods like Support Vector Machines and Random Forests, as well as advanced deep learning (DL). It focuses on the integration of ML into oncological research, particularly its application in tumor progression through the tumor microenvironment, genetic data, histopathological data, and radiological data. This work provides a critical analysis of the challenges associated with the reliability and accuracy of ML models, which limit their clinical integration.

This review offers expert insights and strategies to address these challenges in order to improve the robustness and applicability of ML in real-world oncology settings. By emphasizing the potential for personalized cancer treatment and bridging gaps between technology and clinical needs, this review serves as a comprehensive resource for advancing the integration of ML models into clinical oncology.

Genomic prediction (GP) aims to predict the breeding values of multiple complex traits, usually assumed to be multivariate normally distributed by the largely used statistical methods, thus imposing linear genetic relationships between traits. Although these methods are valuable for GP, they do not account for potential nonlinear genetic relationships between traits in scenarios. For individual traits, this oversight may minimally affect prediction accuracy, but it can limit genetic progress when selection involves multiple traits. Deep learning (DL) offers a promising alternative for capturing nonlinear genetic relationships due to its ability to identify complex patterns without prior assumptions about the data structure. We proposed a novel hybrid model that that combines both DL and GBLUP (DLGBLUP), which uses the output of the traditional GBLUP, and enhances its predicted genetic values (PGV) by accounting for nonlinear genetic relationships between traits using DL. We simulated data with linear and nonlinear genetic relationships between traits in order to verify whether DLGBLUP was able to identify nonlinearity when present and avoid inducing it when absent. We found that DLGBLUP consistently provided more accurate PGV for traits simulated with strong nonlinear genetic relationships, accurately identifying these relationships. Over 7 generations of selection, a greater genetic progress was achieved with PGV that accounted for nonlinear relationships (DLGBLUP), compared with GBLUP. When applied to a real dataset from the French Holstein dairy cattle population, DLGBLUP detected nonlinear genetic relationships between pairs of traits, such as conception rate and protein content, and SCC and fat yield, although, no significant increase in prediction accuracy was observed. The integration of DL into GP enabled the modeling of nonlinear genetic relationships between traits, a possibility not previously discussed, given the linear nature of GBLUP. The detection of nonlinear genetic relationships between traits in the French Holstein population when using DLGBLUP indicates the presence of such relationships in real breeding data, suggesting that it may be relevant to further explore nonlinear relationships. This possibility of nonlinear genetic relationships between traits offers a different perspective into multitrait evaluations, with potential to further improve selection strategies in commercial livestock breeding programs. This is particularly relevant when integrating new traits into multitrait evaluations or incorporating new subpopulations, which may introduce different forms of nonlinearity. Finally, it is shown that DL can be used as a complement to the statistical methods deployed in routine genetic evaluations, rather than as an alternative, by enhancing their performance.

Breast cancer continues to be a major health concern, and early detection is vital for enhancing survival rates. Magnetic resonance imaging (MRI) is a key tool due to its substantial sensitivity for invasive breast cancers. Computer-aided detection (CADe) systems enhance the effectiveness of MRI by identifying potential lesions, aiding radiologists in focusing on areas of interest, extracting quantitative features, and integrating with computer-aided diagnosis (CADx) pipelines. This review aims to provide a comprehensive overview of the current state of CADe systems in breast MRI, focusing on the technical details of pipelines and segmentation models including classical intensity-based methods, supervised and unsupervised machine learning (ML) approaches, and the latest deep learning (DL) architectures. It highlights recent advancements from traditional algorithms to sophisticated DL models such as U-Nets, emphasizing CADe implementation of multi-parametric MRI acquisitions. Despite these advancements, CADe systems face challenges like variable false-positive and negative rates, complexity in interpreting extensive imaging data, variability in system performance, and lack of large-scale studies and multicentric models, limiting the generalizability and suitability for clinical implementation. Technical issues, including image artefacts and the need for reproducible and explainable detection algorithms, remain significant hurdles. Future directions emphasize developing more robust and generalizable algorithms, integrating explainable AI to improve transparency and trust among clinicians, developing multi-purpose AI systems, and incorporating large language models to enhance diagnostic reporting and patient management. Additionally, efforts to standardize and streamline MRI protocols aim to increase accessibility and reduce costs, optimizing the use of CADe systems in clinical practice. LEVEL OF EVIDENCE: NA TECHNICAL EFFICACY: Stage 2.

Acrylamide (AA), a food hazard generated during thermal processing, poses significant safety risks due to its toxicity. Conventional methods for AA toxicology are time-consuming and inadequate for analyzing cellular morphology. This study developed a novel approach combining deep learning models (U-Net and ResNet34) with cell fluorescence imaging. U-Net was used for cell segmentation, generating a single-cell dataset, while ResNet34 trained the dataset over 200 epochs, achieving an 80 % validation accuracy. This method predicts AA concentration ranges by matching cell fluorescence features with the dataset and analyzes cellular phenotypic changes under AA exposure using k-means clustering and CellProfiler. The approach overcomes the limitations of traditional toxicological methods, offering a direct link between cell phenotypes and hazard toxicology. It provides a high-throughput, accurate solution to evaluate AA toxicology and refines the understanding of its cellular impacts.

Photoacoustic spectroscopy (PAS) gas detection is frequently compromised by noise-induced correlation degradation, which significantly impacts measurement accuracy. To mitigate this issue, an advanced convolutional neural network (CNN) architecture, termed PSO-EAP-CNN, is proposed, which combines particle swarm optimization (PSO) with an ensemble augmented prediction (EAP) strategy. The proposed framework employs a multi-scale feature extraction mechanism through its convolutional architecture, while simultaneously optimizing network parameters via PSO, thereby achieving accelerated convergence and improved prediction stability. The incorporation of the EAP strategy further enhances the model's robustness and generalization ability under noisy conditions. Experimental results demonstrate significant improvements: compared to baseline CNN, PSO-EAP-CNN reduces MAE by 43.76 %, RMSE by 39.25 %, and MAPE by 51.15 %; compared to ordinary least squares regression, improvements reach 68.55 %, 67.43 %, and 75.21 % respectively. The model runs in only 10 seconds per execution. This work advances PAS-based gas detection, offering enhanced accuracy and noise resilience for practical trace gas analysis.

The surge in medical waste has highlighted the urgent need for cost-effective and advanced management solutions. In this paper, a novel medical waste management approach, "MedBin," is proposed for automated sorting, reusing, and recycling. A comprehensive medical waste dataset, "MedBin-Dataset" is established, comprising 2,119 original images spanning 36 categories, with samples captured in various backgrounds. The lightweight "MedBin-Net" model is introduced to enable detection and instance segmentation of medical waste, enhancing waste recognition capabilities. Experimental results demonstrate the effectiveness of the proposed approach, achieving an average precision of 0.91, recall of 0.97, and F1-score of 0.94 across all categories with just 2.51 M parameters (where M stands for million, i.e., 2.51 million parameters), 5.20G FLOPs (where G stands for billion, i.e., 5.20 billion floating-point operations per second), and 0.60 ms inference time. Additionally, the proposed method includes a World Health Organization (WHO) Guideline-Based Classifier that categorizes detected waste into 5 types, each with a corresponding disposal method, following WHO medical waste classification standards. The proposed method, along with the dedicated dataset, offers a promising solution that supports sustainable medical waste management and other related applications. To access the MedBin-Dataset samples, please visit https://universe.roboflow.com/uob-ylti8/medbin_dataset. The source code for MedBin-Net can be found at https://github.com/Wayne3918/MedbinNet.

Motion sickness has been a key factor affecting user experience in Virtual Reality (VR) and limiting the development of the VR industry. Accurate detection of Virtual Reality Motion Sickness (VRMS) is a prerequisite for solving the problem.

In this paper, a dual-pathway model with channel attention for detecting VRMS is proposed. The proposed model has two pathways that both consist of CNN blocks and channel attention modules. The first pathway takes the EEG signal as inputs. The second pathway transforms the EEG signal into brain networks of six frequency bands using Phase Locking Value (PLV) or ρ index (RHO) methods and takes the adjacent matrixes as input. The features from the two pathways are connected and fed into the fully connected layer for classification. Finally, a VR flight simulation experiment is performed and the EEG of the resting state before and after the virtual flight task are collected to validate the model.

The average accuracy, precision, recall, and F1 score of the proposed model are 99.12 %, 99.12 %, 99.11 %, and 99.12 %, respectively.

Eight models are introduced as the reference methods and four of them are fused as the hybrid models in this study. The results show that the proposed model is better than those state-of-art models.

The proposed model outperforms the state-of-the-art models and provides objective and direct guidance for overcoming VRMS and optimizing VR experience.

Accurate and secure classifying informal documents related to mental disorders is challenging due to factors such as informal language, noisy data, cultural differences, personal information and mixed emotions. Conventional deep learning models often struggle to capture patterns in informal text, as they miss long-range dependencies, explain words and phrases literally, and have difficulty processing non-standard inputs like emojis. To address these limitations, we expand data dimensionality, transforming and fusing textual data and signs from a 1D to a 2D space. This enables the use of pre-trained 2D CNN models, such as AlexNet, Restnet-50, and VGG-16 removing the need to design and train new models from scratch. We apply this approach to a dataset of social media posts to classify informal documents as either related to suicide or non-suicide content. Our results demonstrate high classification accuracy, exceeding 99%. In addition, our 2D visual data representation conceals individual private information and helps explainability.

The aim of this prospective study was to determine the effectiveness of screening using image processing analysis and a deep convolutional neural network (DCNN) to classify oral cancers using non-invasive fluorescence visualization. The study included 1076 patients with diseases of the oral mucosa (oral cancer, oral potentially malignant disorders (OPMDs), benign disease) or normal mucosa. For oral cancer, the rate of fluorescence visualization loss (FVL) was 96.9%. Regarding image processing, multivariate analysis identified FVL, the coefficient of variation of the G value (CV), and the G value ratio (VRatio) as factors significantly associated with oral cancer detection. The sensitivity and specificity for detecting oral cancer were 96.9% and 77.3% for FVL, 80.8% and 86.4% for CV, and 84.9% and 87.8% for VRatio, respectively. Regarding the performance of the DCNN for image classification, recall was 0.980 for oral cancer, 0.760 for OPMDs, 0.960 for benign disease, and 0.739 for normal mucosa. Precision was 0.803, 0.821, 0.842, and 0.941, respectively. The F-score was 0.883, 0.789, 0.897, and 0.828, respectively. Sensitivity and specificity for detecting oral cancer were 98.0% and 92.7%, respectively. The accuracy for all lesions was 0.851, average recall was 0.860, average precision was 0.852, and average F-score was 0.849.

This study introduces a hybrid approach that combines unsupervised self-organizing maps (SOM) with a supervised convolutional neural network (CNN) to enhance model accuracy in vector-borne disease modeling. We applied this method to predict insecticide resistance (IR) status in key malaria vectors across Africa. Our results show that the combined SOM/CNN approach is more robust than a standalone CNN model, achieving higher overall accuracy and Kappa scores among others. This confirms the potential of the SOM/CNN hybrid as an effective and reliable tool for improving model accuracy in public health applications.•The hybrid model, combining SOM and CNN, was implemented to predict IR status in malaria vectors, providing enhanced accuracy across various validation metrics.•Results indicate a notable improvement in robustness and predictive accuracy over traditional CNN models.•The combined SOM/CNN approach demonstrated higher Kappa scores and overall model accuracy.

Lung cancer remains a leading cause of cancer-related mortality worldwide, emphasizing the critical need for accurate and early diagnostic solutions. This study introduces a novel multimodal artificial intelligence (AI) framework that integrates Convolutional Neural Networks (CNNs) and Artificial Neural Networks (ANNs) to improve lung cancer classification and severity assessment. The CNN model, trained on 1019 preprocessed CT images, classified lung tissue into four histological categories, adenocarcinoma, large cell carcinoma, squamous cell carcinoma, and normal, with a weighted accuracy of 92 %. Interpretability is enhanced using Gradient-weighted Class Activation Mapping (Grad-CAM), which highlights the salient image regions influencing the model's predictions. In parallel, an ANN trained on clinical data from 999 patients-spanning 24 key features such as demographic, symptomatic, and genetic factors-achieves 99 % accuracy in predicting cancer severity (low, medium, high). SHapley Additive exPlanations (SHAP) are employed to provide both global and local interpretability of the ANN model, enabling transparent decision-making. Both models were rigorously validated using k-fold cross-validation to ensure robustness and reduce overfitting. This hybrid approach effectively combines spatial imaging data and structured clinical information, demonstrating strong predictive performance and offering an interpretable and comprehensive AI-based solution for lung cancer diagnosis and management.

Current image-based analysis methods for monitoring cell confluency and status depend on individual interpretations, which can lead to wide variations in the quality of cell therapeutics. To overcome these limitations, images of mesenchymal stem cells cultured adherently in various types of culture vessels were captured and analyzed using a deep neural network. Among the various deep learning methods, a classification and detection algorithm was selected to verify cell confluency and status. We confirmed that the image classification algorithm demonstrates significant accuracy for both single- and multistack images. Abnormal cells could be detected exclusively in single-stack images, as multistack culture was performed only when abnormal cells were absent in the single-stack culture. This study is the first to analyze cell images based on a deep learning method that directly impacts yield and quality, which are important product parameters in stem cell therapeutics.

Mosquitoes, which belong to diverse species, play a significant role in ecological systems and public health. The accurate identification (classification) of mosquito species is essential for a comprehensive understanding of their ecological roles, behaviors, and evolutionary patterns. While numerous studies have attempted to classify the mosquito species based on images, the existing works still have limitations. Our research is focused on vector mosquito classification based on deep ensemble transfer learning. Initially, we employed transfer learning via four pre-trained convolutional neural network (CNN) models. Subsequently, we have proposed the TransembleNet (Transfer Learning-based Ensemble Networks) approach, which is a novel method of generating ensemble learning models using four different combinations of three transfer learning models. All the experiments were done using the Nadam and Adam optimizers, and we have also applied data augmentation techniques. Among the four ensemble models, Ensemble Model 2 (composed of InceptionV3, VGG-16, and ResNet-50) performed the best. It exhibits very high precision, recall, F1-score, and accuracy values on the "Mosquito on Human Skin" dataset by Ong and Ahmed and the "Vector Mosquito" dataset by Park et al. Our proposed method outperformed the state-of-the-art research works for both datasets.

Neurological disorders, ranging from common conditions like Alzheimer's disease that is a progressive neurodegenerative disorder and remains the most common cause of dementia worldwide to rare disorders such as Angelman syndrome, impose a significant global health burden. Altered facial expressions are a common symptom across these disorders, potentially serving as a diagnostic indicator. Deep learning algorithms, especially convolutional neural networks (CNNs), have shown promise in detecting these facial expression changes, aiding in diagnosing and monitoring neurological conditions.

This systematic review and meta-analysis aimed to evaluate the performance of deep learning algorithms in detecting facial expression changes for diagnosing neurological disorders.

Following PRISMA2020 guidelines, we systematically searched PubMed, Scopus, and Web of Science for studies published up to August 2024. Data from 28 studies were extracted, and the quality was assessed using the JBI checklist. A meta-analysis was performed to calculate pooled accuracy estimates. Subgroup analyses were conducted based on neurological disorders, and heterogeneity was evaluated using the I2 statistic.

The meta-analysis included 24 studies from 2019 to 2024, with neurological conditions such as dementia, Bell's palsy, ALS, and Parkinson's disease assessed. The overall pooled accuracy was 89.25% (95% CI 88.75-89.73%). High accuracy was found for dementia (99%) and Bell's palsy (93.7%), while conditions such as ALS and stroke had lower accuracy (73.2%).

Deep learning models, particularly CNNs, show strong potential in detecting facial expression changes for neurological disorders. However, further work is needed to standardize data sets and improve model robustness for motor-related conditions.

Accurate prediction of the turnover number (kcat), which quantifies the maximum rate of substrate conversion at an enzyme's active site, is essential for assessing catalytic efficiency and understanding biochemical reaction mechanisms. Traditional wet-lab measurements of kcat are time-consuming and resource-intensive, making deep learning (DL) methods an appealing alternative. However, existing DL models often overlook the impact of reaction products on kcat due to feedback inhibition, resulting in suboptimal performance. The multimodal nature of this kcat prediction task, involving enzymes, substrates, and products as inputs, presents additional challenges when certain modalities are unavailable during inference due to incomplete data or experimental constraints, leading to the inapplicability of existing DL models. To address these limitations, we introduce MMKcat, a novel framework employing a prior-knowledge-guided missing modality training mechanism, which treats substrates and enzyme sequences as essential inputs while considering other modalities as maskable terms. Moreover, an innovative auxiliary regularizer is incorporated to encourage the learning of informative features from various modal combinations, enabling robust predictions even with incomplete multimodal inputs. We demonstrate the superior performance of MMKcat compared to state-of-the-art methods, including DLKcat, TurNup, UniKP, EITLEM-Kinetic, DLTKcat and GELKcat, using BRENDA and SABIO-RK. Our results show significant improvements under both complete and missing modality scenarios in RMSE, R2, and SRCC metrics, with average improvements of 6.41%, 22.18%, and 8.15%, respectively. Codes are available at https://github.com/ProEcho1/MMKcat.

In this work, we explore the possibility of using the topology and weight distribution of the connectome of a Drosophila, or fruit fly, as a reservoir for multivariate chaotic time-series prediction. Based on the information taken from the recently released full connectome, we create the connectivity matrix of an Echo State Network. Then, we use only the most connected neurons and implement two possible selection criteria, either preserving or breaking the relative proportion of different neuron classes which are also included in the documented connectome, to obtain a computationally convenient reservoir. We then investigate the performance of such architectures and compare them to state-of-the-art reservoirs. The results show that the connectome-based architecture is significantly more resilient to overfitting compared to the standard implementation, particularly in cases already prone to overfitting. To further isolate the role of topology and synaptic weights, hybrid reservoirs with the connectome topology but random synaptic weights and the connectome weights but random topologies are included in the study, demonstrating that both factors play a role in the increased overfitting resilience. Finally, we perform an experiment where the entire connectome is used as a reservoir. Despite the much higher number of trained parameters, the reservoir remains resilient to overfitting and has a lower normalized error, under 2%, at lower regularisation, compared to all other reservoirs trained with higher regularisation.

The emerging flexible iontronic sensing (FITS) technology has introduced a novel modality for tactile perception, mimicking the topological structure of human skin while providing a viable strategy for seamless integration with biological systems. With research progress, FITS has evolved from focusing on performance optimization and structural enhancement to a new phase of integration and intelligence, positioning it as a promising candidate for next-generation wearable devices. Therefore, a review from the perspective of technological development trends is essential to fully understand the current state and future potential of FITS devices. In this review, we examine the latest advancements in FITS. We begin by examining the sensing mechanisms of FITS, summarizing research progress in material selection, structural design, and the fabrication of active and electrode layers, while also analysing the challenges and bottlenecks faced by different segments in this field. Next, integrated systems based on FITS devices are reviewed, highlighting their applications in human-machine interaction, healthcare, and environmental monitoring. Additionally, the integration of artificial intelligence into FITS is explored, focusing on optimizing front-end device design and improving the processing and utilization of back-end data. Finally, building on existing research, future challenges for FITS devices are identified and potential solutions are proposed.

We investigate artificial-intelligence-supported in-line holographic imaging with coherent terahertz (THz) radiation. The goal is to reconstruct three-dimensional (3D) scenes from images obtained with detectors that recorded only the power of the radiation and not the phase. This study proposes a novel approach utilizing a physics-informed deep learning (DL) algorithm to reconstruct objects which partially obscure each other. Taking the angular spectrum theory as prior knowledge, we generate a synthetic dataset consisting of a series of diffraction patterns that contain information about the type of objects to be imaged. This dataset, combined with unlabeled data obtained by experiments, are used for the self-training of a physics-informed neural network (NN). During the training process, the NN iteratively predicts images of the objects from the unlabeled dataset and reincorporates these results back into the training set. This recursive strategy includes experimentally recorded images from the studied object class in the NN training, for which the ground truth is unknown. Furthermore, the approach minimizes mutual interference during object reconstruction, demonstrating its effectiveness even in data-scarce situations. The method has been validated with both simulated and experimental data, showcasing its significant potential to advance the field of 3D THz imaging.

Modeling and visualization of glioma growth could assist in cancer diagnosis, tumor progression prediction, and clinical treatment outcome improvement. However, most studies either failed to make patient-specific predictions or could only display information about tumor size and shape, lacking the capability to characterize the impact of tumor growth on surrounding tissues. In this study, a method (HybrSyn) combining tumor growth model and deformable image registration technique for synthesizing MRIs at arbitrary time point after the detection time has been proposed. Through the tumor growth model, tumor growth process for consecutive time point has been predicted according to the characteristics of tumor cell diffusion and proliferation within the brain. The glioma deformable image registration model was employed to obtain the deformation fields between the tumors at detection time and simulations at subsequent time points. These fields were then mapped to the patient's initial MRI scans to generate the synthetic MRIs corresponding to that time points. To validate the HybrSyn, various experiments were conducted on the BraTS19 and the internal dataset collected from Zigong First People's Hospital. The quantitative results demonstrated a structural similarity of 80.93% between the synthesized MRIs and the patients' MRI scans. The qualitative results indicated that the HybrSyn could effectively capture changes during tumor progression and provide a global view. From the clinical point of view, synthesized longitudinal brain MRIs could potentially aid in presenting the impact of glioma growth on surrounding functional areas, and identifying target regions for personalized treatment planning.

Cardiovascular disease complications are the leading cause of morbidity and mortality in patients with Type 2 diabetes (T2DM). Left ventricular diastolic dysfunction (LVDD) is one of the earliest myocardial characteristics of diabetic cardiac dysfunction. Therefore, we aimed to develop an LVDD-risk predictive model to diagnose cardiac dysfunction before severe cardiovascular complications arise. We trained an artificial neural network model to predict LVDD risk with patients' clinical information. The model showed better performance than classical machine learning methods such as logistic regression, random forest and support vector machine. We further explored LVDD-risk/protective features with interpretability methods in neural network. Finally, we provided a freely accessible web server called LVDD-risk, where users can submit their clinical information to obtain their LVDD-risk probability and the most noteworthy risk indicators.

Gait impairment post-stroke is highly heterogeneous. Prior studies classified heterogeneous gait patterns into subgroups using peak kinematics, kinetics, or spatiotemporal variables. A limitation of this approach is the need to select discrete features in the gait cycle. Using continuous gait cycle data, we accounted for differences in magnitude and timing of kinematics. Here, we propose a machine-learning pipeline combining supervised and unsupervised learning. We first trained a Convolutional Neural Network and a Temporal Convolutional Network to extract features that distinguish impaired from neurotypical gait. Then, we used unsupervised time-series k-means and Gaussian Mixture Models to identify gait clusters. We tested our pipeline using kinematic data of 28 neurotypical and 39 individuals post-stroke. We assessed differences between clusters using ANOVA. We identified two neurotypical gait clusters (C1, C2). C1: normative gait pattern. C2: shorter stride time. We observed three post-stroke gait clusters (S1, S2, S3). S1: mild impairment and increased bilateral knee flexion during loading response. S2: moderate impairment, slow speed, short steps, increased knee flexion during stance bilaterally, and reduced paretic knee flexion during swing. S3: mild impairment, asymmetric swing time, increased ankle abduction during the gait cycle, and reduced dorsiflexion bilaterally. Our results indicate that joint kinematics post-stroke are mostly distinct from controls, and highlight kinematic impairments in the non-paretic limb. The post-stroke clusters showed distinct impairments that would require different interventions, providing additional information for clinicians about rehabilitation targets.

Stochastic Computing has attracted extensive attention in the deployment of neural networks at the edge due to its low hardware cost and high fault tolerance. However, traditional stochastic computing requires a long random bit stream to achieve sufficient numerical precision. The long bit stream, in turn, increases the network inference time, hardware cost, and power consumption, which limits its application in executing tasks such as handwritten recognition, speech recognition, image processing, and image classification at the near-sensor end. To realize high-energy-efficiency and low-cost hardware neural networks at the near-sensor end, a hardware optimization design of convolutional neural networks based on the hybrid encoding of deterministic encoding and binary encoding is proposed. By transforming the output signals from the sensor into deterministic encoding and co-optimizing the network training process, a low-cost and high-energy-efficiency convolution operation network is achieved with a shorter bit stream input. This network can achieve good recognition performance with an extremely short bit stream, significantly reducing the system's latency and energy consumption. Compared with traditional stochastic computing networks, this network shortens the bit stream length by 64 times without affecting the recognition rate, achieving a recognition rate of 99% with a 2-bit input. Compared with the traditional 2-bit stochastic computing scheme, the area is reduced by 44.98%, the power consumption is reduced by 60.47%, and the energy efficiency is increased by 12 times. Compared with the traditional 256-bit stochastic computing scheme, the area is reduced by 82.87%, and the energy efficiency is increased by 1947 times. These comparative results demonstrate that this work has significant advantages in executing tasks such as image classification at the near-sensor end and edge devices.

Cardiovascular disease (CVD) is the leading cause of death worldwide, with a noticeable decline in the approval of new therapeutic interventions. Currently, there is no gold standard for developing new therapies for CVDs, and preclinical models do not translate to clinical efficacy. Therefore, there is an urgent need for in vitro models that more accurately mimic human disease processes. Here we describe a model of the artery consisting of monocultures of human coronary artery endothelial cells (HCAECs) or cocultures of HCAECs with human coronary artery smooth muscle cells (HCASMCs). The model was established in the OrganoPlate® 2-lane-48 UF, a novel microfluidic device, comprised of a microtiter plate footprint with 48 chips. Fluid is circulated in a unidirectional manner by interval rocking. The creation of an air-liquid interface at the inlets at a given inclination is used to select flow paths and establish flow in one direction only, whilst capillary forces ensure the channel remains filled with fluid. We investigated the impact of unidirectional or bidirectional flow conditions. Under unidirectional flow, endothelial cells aligned with the flow direction, decreased fibronectin deposition, and smooth muscle cells presented a non-contractile phenotype, emulating the characteristics of healthy arteries. Contrarily, bidirectional flow mimicked features of early endothelial dysfunction, such as contractile morphology of vessels and increased fibronectin secretion, ICAM-1 staining, and lipid deposits. Vascular inflammation could be induced by the addition of TNFα and IL-1β in both flow conditions. Overall, the OrganoPlate® 2-lane-48 UF is a powerful platform providing both throughput and improved flow control, for creating more physiological models. Its ability to replicate key features of a healthy and diseased artery, its potential use in drug screening, and its compatibility with lab automation make it an invaluable tool for researchers aiming for more accurate and efficient therapeutic development in CVD.

Dental caries is one of the most prevalent chronic diseases among both children and adults, despite being largely preventable. This condition has significant negative impacts on human health and imposes a substantial economic burden. In recent years, scientists and dentists have increasingly started to utilize artificial intelligence (AI), particularly machine learning, to improve the efficiency of dental caries management. This study aims to provide an overview of the current knowledge about the AI-enabled approaches for dental caries management within the framework of personalized patient care. Generally, AI works as a promising tool that can be used by both dental professionals and patients. For dental professionals, it predicts the risk of dental caries by analyzing dental caries risk and protective factors, enabling to formulate personalized preventive measures. AI, especially those based on machine learning and deep learning, can also analyze images to detect signs of dental caries, assist in developing treatment plans, and help to make a risk assessment for pulp exposure during treatment. AI-powered tools can also be used to train dental students through simulations and virtual case studies, allowing them to practice and refine their clinical skills in a risk-free environment. Additionally, AI tracks brushing patterns and provides feedback to improve oral hygiene practices of the patients and the general population, thereby improving their understanding and compliance. This capability of AI can inform future research and the development of new strategies for dental caries management and control.

Intrahepatic cholangiocarcinoma (iCCA) is associated with a poor prognosis and necessitates a multimodal, multidisciplinary approach from diagnosis to treatment to achieve optimal outcomes. A noninvasive preoperative diagnosis using abdominal imaging techniques can represent a clinical challenge. Given the differential response of iCCA to localized and systemic therapies compared with hepatocellular carcinoma and secondary hepatic malignancies, an accurate diagnosis is crucial. Deep learning (DL) models for image analysis have emerged as a promising adjunct for the abdominal radiologist, potentially enhancing the accurate detection and diagnosis of iCCA. Over the last five years, several reports have proposed robust DL models, which demonstrate a diagnostic accuracy that is either comparable to or surpasses that of radiologists with varying levels of experience. Recent studies have expanded DL applications into other aspects of iCCA management, including histopathologic diagnosis, the prediction of histopathological features, the preoperative prediction of survival, and the pretreatment prediction of responses to systemic therapy. We herein critically evaluate the expanding body of research on DL applications in the diagnosis and management of iCCA, providing insights into the current progress and future research directions. We comprehensively synthesize the performance and limitations of DL models in iCCA research, identifying key challenges that serve as a translational reference for clinicians.

Pneumonia is the leading cause of hospitalisation and mortality among children under five, particularly in low-resource settings. Accurate differentiation between viral and bacterial pneumonia is essential for guiding appropriate treatment, yet it remains challenging due to overlapping clinical and radiographic features. Advances in machine learning (ML), particularly deep learning (DL), have shown promise in classifying pneumonia using chest X-ray (CXR) images. This scoping review summarises the evidence on ML techniques for classifying viral and bacterial pneumonia using CXR images in paediatric patients.

This scoping review was conducted following the Joanna Briggs Institute methodology and the PRISMA-ScR guidelines. A comprehensive search was performed in PubMed, Embase, and Scopus to identify studies involving children (0-18 years) with pneumonia diagnosed through CXR, using ML models for binary or multiclass classification. Data extraction included ML models, dataset characteristics, and performance metrics.

A total of 35 studies, published between 2018 and 2025, were included in this review. Of these, 31 studies used the publicly available Kermany dataset, raising concerns about overfitting and limited generalisability to broader, real-world clinical populations. Most studies (n=33) used convolutional neural networks (CNNs) for pneumonia classification. While many models demonstrated promising performance, significant variability was observed due to differences in methodologies, dataset sizes, and validation strategies, complicating direct comparisons. For binary classification (viral vs bacterial pneumonia), a median accuracy of 92.3% (range: 80.8% to 97.9%) was reported. For multiclass classification (healthy, viral pneumonia, and bacterial pneumonia), the median accuracy was 91.8% (range: 76.8% to 99.7%).

Current evidence is constrained by a predominant reliance on a single dataset and variability in methodologies, which limit the generalisability and clinical applicability of findings. To address these limitations, future research should focus on developing diverse and representative datasets while adhering to standardised reporting guidelines. Such efforts are essential to improve the reliability, reproducibility, and translational potential of machine learning models in clinical settings.

Reduced fetal movement (RFM) can indicate that a fetus is at risk, but current monitoring methods provide only a "snapshot in time" of fetal health and require trained clinicians in clinical settings. To improve antenatal care, there is a need for continuous, objective fetal movement monitoring systems. Wearable sensors, like inertial measurement units (IMUs), offer a promising data-driven solution, but distinguishing fetal movements from maternal movements remains challenging. The potential benefits of using linear acceleration and angular rate data for fetal movement detection have not been fully explored. In this study, machine learning models were developed using linear acceleration and angular rate data from twenty-three participants who wore four abdominal IMUs and one chest reference while indicating perceived fetal movements with a handheld button. Random forest (RF), bi-directional long short-term memory (BiLSTM), and convolutional neural network (CNN) models were trained using hand-engineered features, time series data, and time-frequency spectrograms, respectively. The results showed that combining accelerometer and gyroscope data improved detection performance across all models compared to either one alone. CNN consistently outperformed other models but required larger datasets. RF and BiLSTM, while more sensitive to signal noise, offered reasonable performance with smaller datasets and greater interpretability.

Worldwide, 30%-45% of adults have sleep disorders, which are linked to major health issues such as diabetes and cardiovascular disease. Long-term monitoring with traditional in-lab testing is impractical due to high costs. Wearable artificial intelligence (AI)-powered solutions offer accessible, scalable, and continuous monitoring, improving the identification and treatment of sleep problems.

This scoping review aims to provide an overview of AI-powered wearable devices used for sleep disorders, focusing on study characteristics, wearable technology features, and AI methodologies for detection and analysis.

Seven electronic databases (MEDLINE, PsycINFO, Embase, IEEE Xplore, ACM Digital Library, Google Scholar, and Scopus) were searched for peer-reviewed literature published before March 2024. Keywords were selected based on 3 domains: sleep disorders, AI, and wearable devices. The primary selection criterion was the inclusion of studies that utilized AI algorithms to detect or predict various sleep disorders using data from wearable devices. Study selection was conducted in 2 steps: first, by reviewing titles and abstracts, followed by full-text screening. Two reviewers independently conducted study selection and data extraction, resolving discrepancies by consensus. The extracted data were synthesized using a narrative approach.

The initial search yielded 615 articles, of which 46 met the eligibility criteria and were included in the final analysis. The majority of studies focused on sleep apnea. Wearable AI was widely deployed for diagnosing and screening disorders; however, none of the studies used it for treatment. Commercial devices were the most commonly used type of wearable technology, appearing in 30 out of 46 (65%) studies. Among these, various brands were utilized rather than a single large, well-known brand; 19 (41%) studies used wrist-worn devices. Respiratory data were used by 25 of 46 (54%) studies as the primary data for model development, followed by heart rate (22/46, 48%) and body movement (17/46, 37%). The most popular algorithm was the convolutional neural network, adopted by 17 of 46 (37%) studies, followed by random forest (14/46, 30%) and support vector machines (12/46, 26%).

Wearable AI technology offers promising solutions for sleep disorders. These devices can be used for screening and diagnosis; however, research on wearable technology for sleep disorders other than sleep apnea remains limited. To statistically synthesize performance and efficacy results, more reviews are needed. Technology companies should prioritize advancements such as deep learning algorithms and invest in wearable AI for treating sleep disorders, given its potential. Further research is necessary to validate machine learning techniques using clinical data from wearable devices and to develop useful analytics for data collection, monitoring, prediction, classification, and recommendation in the context of sleep disorders.

Natural products have applications as biopharmaceuticals, agrochemicals, and other high-value chemicals. However, there are challenges in isolating natural products from their native producers (e.g. bacteria, fungi, plants). In many cases, synthetic chemistry or heterologous expression must be used to access these important molecules. The biosynthetic machinery to generate these compounds is found within biosynthetic gene clusters, primarily consisting of the enzymes that biosynthesise a range of natural product classes (including, but not limited to ribosomal and nonribosomal peptides, polyketides, and terpenoids). Cell-free synthetic biology has emerged in recent years as a bottom-up technology applied towards both prototyping pathways and producing molecules. Recently, it has been applied to natural products, both to characterise biosynthetic pathways and produce new metabolites. This review discusses the core biochemistry of cell-free synthetic biology applied to metabolite production and critiques its advantages and disadvantages compared to whole cell and/or chemical production routes. Specifically, we review the advances in cell-free biosynthesis of ribosomal peptides, analyse the rapid prototyping of natural product biosynthetic enzymes and pathways, highlight advances in novel antimicrobial discovery, and discuss the rising use of cell-free technologies in industrial biotechnology and synthetic biology.

Developmental dysplasia of the hip (DDH) is a common neonatal condition that necessitates early diagnosis to ensure effective treatment. The traditional Graf method, while widely used for evaluating infant hips via ultrasound, is limited by operator dependency and measurement variability. This research has proposed a framework using deep learning network, morphological operation and local maxima method to diagnose DDH in newborns using ultrasound images. The method utilizes DeepLabv3 + for image segmentation, evaluating multiple backbone architectures (ResNet50, InceptionResNetV2, MobilenetV2, and Xception) to identify the region of interest accurately. Local maxima method was used to determine the extremum points of the line defining the Graf angles. Denoising filters, including mean, median, and Wiener, are applied to determine local maxima points accurately. The evaluation comprises two stages: first, assessing the performance of DeepLabv3 + backbones in producing masks for Graf angles determination, and second, comparing the angles obtained through proposed framework with those determined by expert radiologists. Comparative analysis demonstrates that MobileNetV2 (94.64 accuracy, 86.99 Cohen's kappa, 94.31 F-score) surpasses other models in segmentation accuracy and measurement reliability. This conclusion is backed by key performance metrics such as accuracy, IoU, PSNR, F-score, SSIM, Cohen's kappa, as well as by the intraclass correlation coefficient and Bland-Altman analyses. The proposed framework shows considerable promise in automating hip ultrasound analysis for DDH diagnosis, minimizing operator dependency while enhancing measurement consistency.

Most microbiota determination (skin, gut, soil, etc.) are currently conducted in a laboratory using expensive equipment and lengthy procedures, including culture-dependent methods, nucleic acid amplifications (including quantitative PCR), DNA microarray, immunoassays, 16S rRNA sequencing, shotgun metagenomics, and sophisticated mass spectrometric methods. In situ and rapid analysis methods are desirable for fast turnaround time and low assay cost. Fluorescence identification of bacteria and their mixtures is emerging to meet this demand, thanks to the recent development in various machine learning methods. High-dimensional spectroscopic or microscopic imaging data can be obtained to identify the bacterial makeup and its implications for human health and the environment. For example, we can classify healthy versus non-healthy skin microbiome, inflammatory versus non-inflammatory gut microbiome, degraded versus non-degraded soil microbiome, etc. This tutorial summarizes the various machine-learning algorithms used in bacteria identification and microbiota determinations. It also summarizes the various fluorescence spectroscopic methods used to identify bacteria and their mixtures, including fluorescence lifetime spectroscopy, fluorescence resonance energy transfer (FRET), and synchronous fluorescence (SF) spectroscopy. Finally, various fluorescence microscopic imaging methods were summarized that have been used to identify bacteria and their mixtures, including epi-fluorescence microscopy, confocal microscopy, two-photon/multi-photon microscopy, and super-resolution imaging methods (STED, SIM, PALM, and STORM). Finally, it discusses how these methods can be applied to microbiota determinations, what can be demonstrated in the future, opportunities and challenges, and future directions.

Recent achievements in the fields of deep learning and remote sensing have led to their application in monitoring river water quality. One of the most researched methods is the estimation of total suspended solid (TSS) concentrations using multispectral imagery and convolutional neural network (CNN) models. Owing to the sorption capacity of other pollutants, TSS monitoring is essential. However, despite recent advances in deep learning, the application of contemporary technologies in water quality monitoring has not yet been fully explored. This study aims to develop a framework for on-device AI that can be applied to edge devices through quantization using a lightweight deep learning model. Lightweight CNN models were identified using neural architecture search (NAS) in conjunction with Pareto optimization, achieving high performance (0.806 of Nash-Sutcliffe efficiency (NSE)) while minimizing computational burden (8.118 MB). The model sizes were further compressed (0.736 MB) through the application of post-training quantization (PTQ) and quantization aware training (QAT), ensuring that accuracy (0.831 of NSE) was preserved. This provides a scalable approach for real-time TSS monitoring, bridging the gap between advanced deep learning techniques and practical environmental applications. These applications indicate that it is possible to estimate other water quality indices using multispectral imagery. It enables the tracing of the source of contamination and facilitates rapid responses by identifying changes in real time.

In this study, we fabricated SiOx-based interface-type resistive random-access memory (ReRAM) devices and demonstrated their superior performance. The device was operated at voltages below 3 V with a maximum current of less than 1 mA. It exhibited an on/off ratio of approximately 10, with a set speed of 1 μs at 3 V and a reset speed of 1 μs at -4.5 V. Notably, the retention time at 85 °C reached 104 s. The interface-type ReRAM displayed significant linearity owing to gradual operation, which is characteristic of long-term potentiation and long-term depression. This high linearity facilitates an impressive modified national institute of standards and technology database (MNIST) digit recognition accuracy of 92.21%. To further understand the influence of endurance on learning performance, we evaluated the impact of synaptic weight degradation by comparing both TiN/SiOx/TiN and Pt/SiOx/Pt devices. This approach allowed us to assess how degradation directly affects synaptic behavior and learning efficiency in neuromorphic applications. The TiN/SiOx/TiN configuration exhibited superior endurance, as the presence of an oxygen reservoir improved synaptic performance and stability, which aligns with the gradual switching dynamics observed in our experiments and contributes to the overall device robustness and efficiency.

Artificial intelligence (AI) has rapidly advanced and shows great potential in the prediction, diagnosis, treatment, and prognosis of fatty liver disease (FLD). This study aims to summarize AI's applications and emerging trends in FLD to inspire future research directions.

We analyzed 270 articles sourced from the Web of Science Core Collection published between 2006 and 2024. The study focuses on the medical application of AI in FLD, examining the contributions of authors, institutions, countries, keywords, and cited references.

AI is predominantly applied in FLD diagnosis, with progression from simple diagnostic tools to advanced methods for classifying FLD and assessing liver fat content. Moreover, the types of data used in AI development have evolved, incorporating a variety of new image and clinical data sources. AI is also being integrated into drug development and personalized nutritional therapies for FLD. Additionally, researchers are becoming increasingly interested in the application of AI to study FLD genes.

We found that the applications of AI in FLD are mainly reflected in the prediction, diagnosis, therapy, and prognosis of FLD. In contrast to traditional medicine, AI has the potential to advance the fields of precision medicine and telemedicine, as well as to conserve additional social resources. Moreover, AI may help medical personnel from the perspective of traditional Chinese medicine, FLD prognosis, and the use of AI to analyze gene prediction and natural language processing (NLP).

Convolutional neural networks (CNNs) are increasingly used to classify medical images, but few studies utilize smartphone photographs. The objective of this study was to assess CNNs for differentiating patients from controls and detecting joint inflammation.

We included consecutive patients with early inflammatory arthritis and healthy controls, all examined by a rheumatologist (15% by two). Standardized hand photographs of the hands were taken, anonymized and cropped around joints. Pre-trained CNN models were fine-tuned on our dataset (80% training; 20% test set). We used an Inception-ResNet-v2 backbone CNN modified for two class outputs (patient vs control) on uncropped photos. Separate Inception-ResNet-v2 CNNs were trained on cropped photos of middle finger proximal interphalangeal (MFPIP), index finger proximal interphalangeal (IFPIP) and wrist. We report accuracy, sensitivity, specificity and area under the receiver operating characteristic curve (AUC).

We analysed 800 hands from 200 controls (mean age 37.8 years) and 200 patients (mean age 49 years). Two rheumatologists showed 0.89 concordance. The wrist was commonly involved (173/400) followed by the MFPIP (134) and IFPIP (128). The screening CNN achieved 99% accuracy and specificity and 98% sensitivity in predicting a patient compared with controls. Joint-specific CNN accuracy, sensitivity, specificity and AUC were as follows: wrist (75%, 92%, 72% and 0.86, respectively), IFPIP (73%, 89%, 72% and 0.88, respectively) and MFPIP (71%, 91%, 70% and 0.87, respectively).

Computer vision distinguishes patients and controls using smartphone photographs, showing promise as a screening tool. Future research will focus on validating findings in diverse populations and other joints and integrating this technology into clinical workflows.