Training Dataset Research Articles

The Effect of Dataset Imbalance on the Performance of Image-to-Cartoon Generative Adversarial Networks

This report investigates the impact of dataset imbalance on AI-powered image-to-anime style transfer, focusing on the AnimeGANv3 model. Despite the common perception of AI as free of human bias, we highlight that machine learning systems inherently reflect societal prejudices through their training data. The anime art style, popular worldwide but limited in its representation of diverse ethnicities and cultures, serves as a case study for this phenomenon. We analysed AnimeGANv3's training datasets and compared its performance on over- and underrepresented image classes using quantitative and qualitative metrics. Results demonstrate that users from minority groups likely experience inferior outcomes due to dataset imbalance. The study emphasises the need for transparent and responsible dataset curation for machine learning systems to ensure ethical AI development and improved model performance across all user groups.

Mapping fine-scale socioeconomic inequality using machine learning and remotely sensed data

Abstract Limited and missing socioeconomic data have made it nearly impossible to measure or estimate inequality consistently at fine spatiotemporal and jurisdictional scales, especially for lower- and middle-income countries (L&MICs). We deploy a novel data harmonization method that combines existing household survey data with freely available remotely sensed data and machine learning techniques to generate fine-scale socioeconomic inequality measures across spatial and temporal scales for India. Our manuscript makes three important contributions. One, it identifies key remote sensing datasets that, in combination with nighttime luminosity, improve its predictive power to estimate measures of socioeconomic inequality; Two, it offers an analytical approach that reliably estimates the uneven distribution of socioeconomic conditions by harmonizing household assets and socio-demographic information that remotely sensed data at the village or similar geographic levels represent – the results achieve > 84% prediction accuracy. Finally, it leverages a spatially cross-validated machine learning model with training and test datasets from two successive Demographic & Health Surveys (DHS) to demonstrate how data gaps in socioeconomic inequality at subnational levels can be addressed. Our replicable approach has the potential to improve global inequality data, thereby supporting research and applications aiming to reduce socioeconomic inequality in the context of the Sustainable Development Goals (SDGs).

Development and validation of a novel bleeding risk prediction tool for aspirin users with a low body mass index

Aspirin is commonly utilized in the management and prevention of various diseases. However, in specific individuals, particularly those with low body mass index (BMI), aspirin can elevate the risk of bleeding. Achieving a delicate equilibrium between the desirable antiplatelet effects and potential bleeding complications is a notable consideration. The objective of this study was to create a novel bleeding risk prediction tool for aspirin users with a low BMI. A total of 2436 aspirin users with a low BMI were included in this study conducted at the Affiliated Dongyang Hospital of Wenzhou Medical University. Patient data, comprising demographics, clinical characteristics, comorbidities, medical history, and laboratory tests, were collected. The patients were randomly divided into two groups, with a 7:3 ratio, for model development and internal validation purposes. The identification of clinically significant features associated with bleeding was achieved through the utilization of the Least Absolute Shrinkage and Selection Operator (LASSO) regression and boruta analysis. Subsequently, these important features underwent multivariate logistic regression analysis. Based on independent bleeding risk factors, a logistic regression model was constructed and presented as a nomogram. Model performance was evaluated using metrics such as the area under the curve (AUC), calibration curves, decision curve analysis (DCA), clinical impact curve (CIC), and net reduction curve (NRC) in both the training and testing sets. LASSO analysis identified two clinical features, while Boruta analysis identified nine clinical features out of a total of 21 features. Subsequent multivariate logistic regression analysis selected significant independent risk factors. The boruta model, which demonstrated the highest AUC, consisted of six clinical variables: hemoglobin, platelet count, previous bleeding, tumor, smoke, and diabetes mellitus. These variables were integrated into a visually represented nomogram. The model exhibited an AUC of 0.832 (95% CI: 0.788–0.875) in the training dataset and 0.775 (95% CI: 0.698–0.853) in the test dataset, indicating excellent discriminatory performance. Calibration curve analysis revealed close alignment with the ideal curve. Furthermore, DCA, CIC, and NRC demonstrated favorable clinical net benefit for the model. This study has successfully created a novel risk prediction tool specifically designed for aspirin users with a low BMI. This tool enables the stratification of low BMI patients based on their anticipated bleeding risk.

Solution-processed organic/inorganic heterojunction synaptic transistor for neuromorphic computing

Abstract Artificial synaptic devices are the hardware foundation of modern computing systems which have shown great potential in overcoming the bottleneck of traditional von-Neumann computing architectures. Organic synaptic transistors have garnered considerable attention due to their merits, such as low cost, low weight, and mechanical flexibility. Various materials are utilized for the charge-capture layer in organic synaptic transistors. Indium gallium zinc oxide (IGZO) is a typical metal oxide semiconductor with a wide bandgap, high carrier mobility, and stable characteristics. Moreover, IGZO is an n-type semiconductor with a lower HOMO and LUMO energy level compared to p-type semiconductor, which has great potential as a capture material to fabricate high-performance synaptic devices. However, the application of IGZO as the trapping layer in organic synaptic transistors has received limited attention. Consequently, an organic synaptic transistor based on organic/inorganic heterojunction was developed. The impact of program/erase time on memory performance was investigated, revealing that the memory window and memory ratio increased as the write/erase time was extended. Additionally, typical synaptic behavior were successfully emulated, including excitatory/inhibitory postsynaptic current (EPSC/ IPSC), paired-pulse facilitation (PPF), paired-pulse depression (PPD), high-pass filtering characteristics, and the transformation of short-term plasticity (STP) to long-term plasticity (LTP). Notably, the synaptic transistor based on an inorganic-organic bilayer heterojunction achieved a high recognition accuracy of 89.2% using the MNIST dataset for handwritten digit training. This study provides a facile route for fabricating high-performance synaptic transistors, paving the way for the development of advanced brain-like computers.

Intelligent leaching of Zn and Mn from spent disposable batteries to avoid traditional optimizing experiments.

Spent disposable Zn-Mn and Zn-C batteries are important resources for recycling. Acid leaching is the crucial step in the hydrometallurgy process for recycling Zn and Mn from these spent Zn-based batteries. However, to obtain the optimal leaching efficiency, the uncontrollable components in waste feed and various leaching parameters cause numerous replicated optimal experiments, increasing the recovery cost and environmental risks. To solve the issues, we employed machine learning (ML) techniques to construct models to predict Zn and Mn leaching from spent disposable batteries without optimizing experiments. Among four ML algorithms tested, the extreme gradient boosting demonstrated superior predictive performance, achieving an R2 of 0.85-0.98 across the training, test, and verification datasets. An analysis of feature importance indicated that the particle size, waste composition, acid concentration, temperature, and time affected the metal leaching most. This study also revealed the interaction effects of the waste properties and leaching process on the metal leaching. Furthermore, we created a user-friendly graphical user interface (GUI) that enables quick acquisition of metal leaching results, requiring only the measurement of waste particle size and component. Finally, experimental verification confirmed the practicability of the GUI. This study achieves intelligent metal leaching from spent batteries and overcomes the high recovery cost and environmental risks associated with traditional experimental optimizing methods.

Predicting cardiovascular outcomes in Chinese patients with type 2 diabetes by combining risk factor trajectories and machine learning algorithm: a cohort study

BackgroundCardiovascular complications are major concerns for Chinese patients with type 2 diabetes. Accurately predicting these risks remains challenging due to limitations in traditional risk models. We aimed to develop a dynamic prediction model using machine learning and longitudinal trajectories of cardiovascular risk factors to improve prediction accuracy.MethodsWe included 16,378 patients from the Kailuan cohort, splitting them into training and testing datasets. Using baseline characteristics and changes over a four-year observation period, we developed the ML-CVD-C (Machine Learning Cardiovascular Disease in Chinese) score to predict 10-year cardiovascular risk, including cardiovascular death, nonfatal myocardial infarction, and stroke. We compared the discrimination and calibration of ML-CVD-C with models using only baseline variables (ML-CVD-C [base]), China-PAR (Prediction for ASCVD Risk in China), and PREVENT (Predict Risk of cardiovascular disease EVENTs). Risk stratification improvements were assessed through net reclassification improvement (NRI) and integrated discrimination improvement (IDI). Transition analysis examined the changes in risk stratification over time.ResultsThe ML-CVD-C score achieved a C-index of 0.80 (95% CI: 0.78–0.82) in the testing cohort, significantly outperforming the ML-CVD-C (base) score, China-PAR, and PREVENT, which had C-index values of 0.62–0.65. ML-CVD-C also provided more accurate cardiovascular risk estimates, though all models tended to overestimate the prevalence of high-risk cases. Stratification by the ML-CVD-C score showed substantial improvement, with NRI gains of 57.7%, 44.1%, and 47.3%, and IDI gains of 10.1%, 7.9%, and 8.4% compared to the other three scores. Both the trajectory and machine learning algorithm contributed significantly to the enhancement of model performance. Transition analysis revealed that participants who remained in the same risk category or were reclassified to a lower category exhibited 22% and 86% reductions in cardiovascular risk compared to those reclassified to a higher risk category during the observation period.ConclusionsThe ML-CVD-C model, incorporating dynamic cardiovascular risk trajectories and a machine learning algorithm, significantly improves risk prediction accuracy for Chinese patients with diabetes. This model may serve as a valuable tool for more personalized cardiovascular risk management in type 2 diabetes.

Validation of Laboratory Risk Indicator for Necrotizing Fasciitis (LRINEC) score and establishment of novel score in Japanese patients with necrotizing fasciitis (J-LRINEC score).

The Laboratory Risk Indicator for Necrotizing Fasciitis (LRINEC) score is widely used to distinguish between necrotizing fasciitis and cellulitis. However, LRINEC scores are not as sensitive or specific as initially reported, possibly due to differences in patient backgrounds in different countries. Here, we examined the validity of LRINEC scores in Japanese patients. We also investigated the possibility of developing a new scoring system. Patients with necrotizing fasciitis (n = 56) and cellulitis (n = 209) were retrospectively evaluated. The data were split into training (n = 199) and validation (n = 66) datasets. A logistic regression analysis was used to calculate the C-statistics of the LRINEC scores. A new equation was formulated using logistic regression analysis with an appropriate variable selection (Laboratory Risk Indicator for Necrotizing Fasciitis for Japanese Patients [J-LRINEC] score). The J-LRINEC score had a C-statistic of 0.9683, sensitivity of 91.4%, and specificity of 84.8%. The LRINEC score had a C-statistic of 0.914 and specificity of 96%; however, its usefulness was limited by its sensitivity of 68.9%. Our results suggest that the LRINEC score is valid for Japanese patients; however, the J-LRINEC score showed higher sensitivity and specificity, suggesting that it may be a useful tool for differentiating cellulitis from necrotizing fasciitis among Japanese patients.

Exploring the prognostic role of microbial and genetic markers in lung squamous cell carcinoma

Despite advances in diagnostic and therapeutic strategies, the prognosis of lung squamous cell carcinoma (LUSC) patients remains poor, and the potential of microbiome-based prognostic biomarkers and therapeutic targets remains largely unexplored. LUSC patient data from The Cancer Genome Atlas (TCGA), including microbial genus level abundance data and RNA sequencing (RNA-Seq) data, were used as a training dataset. Two other independent datasets GSE19188 and GSE157009 serve as validation datasets. A microbiome-based risk score (RS) model was constructed by univariate Cox regression analysis combined with the least absolute contraction and selection operator (LASSO) regression. 18 microbial genera were found to be significantly associated with RFS in LUSC patients. The microbial signature built with these microbial genera, exhibited robust predictive accuracy in both the training and validation datasets. Furthermore, hub mRNA between high- and low-risk groups were selected by XGBOOST and intersect with mRNAs screened by univariate Cox regression analysis, finally identifying four mRNA significantly associated with LUSC prognosis. This study reveals a complex interplay between the lung microbiome and genetic biomarkers, and identifies specific microbial-based and mRNA associated with prognosis in LUSC. These findings provide a basis for future studies aimed to elucidate the mechanisms underlying these associations and provide potential biomarkers for guiding treatment decisions and improving patient outcomes.

DeepCERES: A deep learning method for cerebellar lobule segmentation using ultra-high resolution multimodal MRI.

This paper introduces a novel multimodal and high-resolution human brain cerebellum lobule segmentation method. Unlike current tools that operate at standard resolution (1 mm3) or using mono-modal data, the proposed method improves cerebellum lobule segmentation through the use of a multimodal and ultra-high resolution (0.125 mm3) training dataset. To develop the method, first, a database of semi-automatically labelled cerebellum lobules was created to train the proposed method with ultra-high resolution T1 and T2 MR images. Then, an ensemble of deep networks has been designed and developed, allowing the proposed method to excel in the complex cerebellum lobule segmentation task, improving precision while being memory efficient. Notably, our approach deviates from the traditional U-Net model by exploring alternative architectures. We have also integrated deep learning with classical machine learning methods incorporating a priori knowledge from multi-atlas segmentation which improved precision and robustness. Finally, a new online pipeline, named DeepCERES, has been developed to make available the proposed method to the scientific community requiring as input only a single T1 MR image at standard resolution.

Innovating bio-lubricant design: utilizing Mordred descriptors for model development and thermal stability prediction of novel compounds

Purpose Developing bio-lubricants for high-temperature applications, such as engine operations, requires reliable thermal stability assessment. This study aims to create a predictive tool to evaluate thermal stability using existing organic compounds data. The model predicts the onset temperature (T onset) of bio-lubricant candidates derived from epoxidation with various amine nucleophiles, enhancing initial selection efficiency and reducing reliance on traditional trial-and-error methods. Design/methodology/approach Using the Python library Mordred, molecular descriptors were calculated from the SMILES structures of 126 compounds with known T onsets. These descriptors were inputs for machine learning models predicting the thermal stability of bio-lubricants. The models were evaluated on training and test data sets and validated with novel synthesized compounds. Findings The predictive model showed strong performance on the test data set, with an R² of 0.93 and a root mean square error of 17.77. In external validation, the model estimated the thermal stability of a novel bio-lubricant compound (MA-EPO) at 290.7°C, closely matching the actual T onset of approximately 300°C. Originality/value This research introduces a method for predicting the thermal stability of bio-lubricants using machine learning and open-source libraries. This approach significantly advances the field by improving the efficiency of bio-lubricant development for high-temperature applications. It provides a cost-effective and time-saving alternative to traditional experimental methods, serving as a valuable resource for researchers and developers in the bio-lubricant industry.

Swarm learning with weak supervision enables automatic breast cancer detection in magnetic resonance imaging

BackgroundOver the next 5 years, new breast cancer screening guidelines recommending magnetic resonance imaging (MRI) for certain patients will significantly increase the volume of imaging data to be analyzed. While this increase poses challenges for radiologists, artificial intelligence (AI) offers potential solutions to manage this workload. However, the development of AI models is often hindered by manual annotation requirements and strict data-sharing regulations between institutions.MethodsIn this study, we present an integrated pipeline combining weakly supervised learning—reducing the need for detailed annotations—with local AI model training via swarm learning (SL), which circumvents centralized data sharing. We utilized three datasets comprising 1372 female bilateral breast MRI exams from institutions in three countries: the United States (US), Switzerland, and the United Kingdom (UK) to train models. These models were then validated on two external datasets consisting of 649 bilateral breast MRI exams from Germany and Greece.ResultsUpon systematically benchmarking various weakly supervised two-dimensional (2D) and three-dimensional (3D) deep learning (DL) methods, we find that the 3D-ResNet-101 demonstrates superior performance. By implementing a real-world SL setup across three international centers, we observe that these collaboratively trained models outperform those trained locally. Even with a smaller dataset, we demonstrate the practical feasibility of deploying SL internationally with on-site data processing, addressing challenges such as data privacy and annotation variability.ConclusionsCombining weakly supervised learning with SL enhances inter-institutional collaboration, improving the utility of distributed datasets for medical AI training without requiring detailed annotations or centralized data sharing.

Predicting cognitive decline: Deep-learning reveals subtle brain changes in pre-MCI stage.

Mild cognitive impairment (MCI) and preclinical MCI (e.g., subjective cognitive decline, SCD) are considered risk states of dementia, such as Alzheimer's Disease (AD). However, it is challenging to accurately predict conversion from normal cognition (NC) to MCI, which is important for early detection and intervention. Since neuropathological changes may have occurred in the brain many years before clinical AD, we sought to detect the subtle brain changes in the pre-MCI stage using a deep-learning method based on structural Magnetic Resonance Imaging (MRI). To discover early structural neuroimaging changes that differentiate between stable and progressive cognitive status, and to establish a predictive model for MCI conversion. We first created a unique deep-learning framework for pre-AD conversion prediction through the Alzheimer's Disease Neuroimaging Initiative-1 (ADNI-1) database (n = 845). Then, we tested the model on ADNI-2 (n = 321, followed 3 years) and our private study (n = 109), the China Longitudinal Aging Study (CLAS), to validate the rationality for pre-MCI conversion prediction. The CLAS is a 7-year community-based cohort study in Shanghai. Our framework consisted of two steps: 1) a single-ROI-based network (SRNet) for identifying informative regions in the brain, and 2) a multi-ROI-based network (MRNet) for pre-AD conversion prediction. We then utilized these "ROI-based deep learning" neural networks to create a composite score using advanced algorithm-building. We coined this score as the Progressive Index (PI), which serves as a metric for assessing the propensity of AD conversion. Ultimately, we employed the PI to gauge its predictive capability for MCI conversion in both ADNI-2 and CLAS datasets. We primarily utilized baseline T1-weighted MRI scans to identify the most discriminative brain regions and subsequently developed the PI in both training and validation datasets. We compared the PI across different cognitive groups and conducted logistic regression models along with their AUCs, adjusting for education level, gender, neuropsychological test scores, and the presence of comorbid conditions. We trained the SRNet and MRNet using 845 subjects from ADNI-1 with baseline MRI data, in which AD and progressive MCI (converting to AD within 3 years) patients were considered as positive samples, while NC and stable MCI (remaining stable for 3 years) subjects were considered as negative samples. The convolutional neural networks identified the top 10 regions of interest (ROIs) for distinguishing progressive from stable cases. These key brain regions included the hippocampus, amygdala, temporal lobe, insula, and anterior cerebellum. A total of 321 subjects from ADNI-2, including 209 NC (18 progressive NC (pNC), 113 stable NC (sNC), and 78 remaining NC (rNC)) and 112 SCD (11 pSCD, 5 sSCD, and 96 rSCD), as well as 109 subjects from CLAS, including 17 sNC, 16 pNC, 52 sSCD and 24 pSCD participated in the test set, separately. We found that the PI score effectively sorted all subjects by their stages (stable vs progressive). Furthermore, the PI score demonstrated excellent discrimination between the two outcomes in the CLAS data(p<0.001), even after controlling for age, gender, education level, depression symptoms, anxiety symptoms, somatic diseases, and baseline MoCA score. Better performance for prediction progression to MCI in CLAS was obtained when the PI score was combined with clinical measures (AUC=0.812; 95 %CI: 0.725-0.900). This study effectively predicted the progression to MCI among order individuals at normal cognition state by deep learning algorithm with MRI scans. Exploring the key brain alterations during the very early stages, specifically the transition from NC to MCI, based on deep learning methods holds significant potential for further research and contributes to a deeper understanding of disease mechanisms.

Comparison of algorithms for the recognition of ChatGPT paraphrased texts

The rapid development of artificial intelligence, especially AI assistants, is leading to new forms of plagiarism that are difficult to detect using existing methods. Paraphrasing tools make this problem even more complex and challenging especially in minor languages with inadequate resources and tools. This study explores strategies to help detect plagiarism generated by ChatGPT 4.0 and altered by paraphrasing tools. We propose two new datasets consisting of abstracts of doctoral theses in English and Serbian. Both datasets were subjected to ChatGPT paraphrasing, which allowed us to form two classes of texts: human-written and AI-generated, i.e., AI-paraphrased. We then comprehensively compare 19 widely used classification algorithms based on two feature sets: word unigrams and character multigrams. In addition, we compare these to the results of a commercially available pre-trained ChatGPT content detector, ZeroGPT. The results on the English corpus turn out to be very accurate, achieving an accuracy of 95% or more. In contrast, the results on the Serbian corpus were less accurate, achieving an accuracy of just over 85%. Syntax analysis of the training datasets has shown that in Serbian GPT-paraphrased texts, 33.2% of sentences remain the same, and they are found in 63% of documents. GPT-paraphrased English texts showed that 3.2% of sentences remain the same, and they are found in 16% of documents. Syntax analysis of the test datasets has shown that the change of the model temperature influences syntactic features (average number of words and sentences) in English texts and slightly or not in Serbian texts. We attribute all these differences to GPT’s lower paraphrasing ability in minor languages such as Serbian. Presented findings underscore the necessity for making persistent effort in developing tools made for detecting AI-paraphrased texts in academic and professional settings, particularly for minor languages with limited NLP resources, to preserve content integrity and authenticity.

Mean Field Initialization of the Annealed Importance Sampling Algorithm for an Efficient Evaluation of the Partition Function Using Restricted Boltzmann Machines

Probabilistic models in physics often require the evaluation of normalized Boltzmann factors, which in turn implies the computation of the partition function Z. Obtaining the exact value of Z, though, becomes a forbiddingly expensive task as the system size increases. A possible way to tackle this problem is to use the Annealed Importance Sampling (AIS) algorithm, which provides a tool to stochastically estimate the partition function of the system. The nature of AIS allows for an efficient and parallel implementation in Restricted Boltzmann Machines (RBMs). In this work, we evaluate the partition function of magnetic spin and spin-like systems mapped into RBMs using AIS. So far, the standard application of the AIS algorithm starts from the uniform probability distribution and uses a large number of Monte Carlo steps to obtain reliable estimations of Z following an annealing process. We show that both the quality of the estimation and the cost of the computation can be significantly improved by using a properly selected mean-field starting probability distribution. We perform a systematic analysis of AIS in both small- and large-sized problems, and compare the results to exact values in problems where these are known. As a result, we propose two successful strategies that work well in all the problems analyzed. We conclude that these are good starting points to estimate the partition function with AIS with a relatively low computational cost. The procedures presented are not linked to any learning process, and therefore do not require a priori knowledge of a training dataset.

A Collection of Machine Learning Models for Improved Airport Operations Amidst Adverse Weather Conditions

In the face of escalating climate change, airports worldwide are finding themselves at the mercy of extreme weather events. This research paper presents a comprehensive system that models key indicators, aiding airport management during such challenging weather conditions. The system adopts an integrated approach, combining various machine learning models to provide a detailed projection of an airport's future state, drawing from past occurrences. The heart of the system is a model that focuses on the airport's peak service rate. This model meticulously correlates weather conditions and runway configurations with the 99th percentile of observed throughput from the training dataset. As such, the peak service rate model provides an estimate of the airport's capacity, which is essential for effective planning and resource allocation. Moreover, the system includes a predictive model that assesses the likelihood of air traffic flow management regulations based on weather data and calendar information. The robustness of this model against noise and uncertainty in the training dataset is fortified by the application of confident learning techniques and the inclusion of monotonic constraints. The system further enhances its capabilities by forecasting the potential entry rate of regulations, expressed in hourly arrivals, providing valuable insights that can guide proactive decision-making. By seamlessly integrating these three models, the system serves as an effective tool for airport operators and airlines. It enables operational optimisation and the development of strategic plans to mitigate the effects of increasing weather-related disruptions.

Mortality Risk Prediction in Patients With Antimelanoma Differentiation-Associated, Gene 5 Antibody-Positive, Dermatomyositis-Associated Interstitial Lung Disease: Algorithm Development and Validation.

Patients with antimelanoma differentiation-associated gene 5 antibody-positive dermatomyositis-associated interstitial lung disease (anti-MDA5+DM-ILD) are susceptible to rapidly progressive interstitial lung disease (RP-ILD) and have a high risk of mortality. There is an urgent need for a reliable prediction model, accessible via an easy-to-use web-based tool, to evaluate the risk of death. This study aimed to develop and validate a risk prediction model of 3-month mortality using machine learning (ML) in a large multicenter cohort of patients with anti-MDA5+DM-ILD in China. In total, 609 consecutive patients with anti-MDA5+DM-ILD were retrospectively enrolled from 6 hospitals across China. Patient demographics and laboratory and clinical parameters were collected on admission. The primary endpoint was 3-month mortality due to all causes. Six ML algorithms (Extreme Gradient Boosting [XGBoost], logistic regression (LR), Light Gradient Boosting Machine [LightGBM], random forest [RF], support vector machine [SVM], and k-nearest neighbor [KNN]) were applied to construct and evaluate the model. After applying inclusion and exclusion criteria, 509 (83.6%) of the 609 patients were included in our study, divided into a training cohort (n=203, 39.9%), an internal validation cohort (n=51, 10%), and 2 external validation cohorts (n=92, 18.1%, and n=163, 32%). ML identified 8 important variables as critical for model construction: RP-ILD, erythrocyte sedimentation rate (ESR), serum albumin (ALB) level, age, C-reactive protein (CRP) level, aspartate aminotransferase (AST) level, lactate dehydrogenase (LDH) level, and the neutrophil-to-lymphocyte ratio (NLR). LR was chosen as the best algorithm for model construction, and the model demonstrated excellent performance, with an area under the receiver operating characteristic (ROC) curve (AUC) of 0.866, a sensitivity of 84.8%, and a specificity of 84.4% on the validation data set and an AUC of 0.90, a sensitivity of 85.0%, and a specificity of 83.9% on the training data set. Calibration curves and decision curve analysis (DCA) confirmed the model's accuracy and clinical applicability. Moreover, the model showed strong predictive performance in the external validation cohorts (cohort 1: AUC=0.836, 95% CI 0.754-0.916; cohort 2: AUC=0.915, 95% CI 0.871-0.959), indicating good generalizability. This model was integrated into a web-based tool to predict the 3-month mortality for patients with anti-MDA5+DM-ILD. We successfully developed a robust clinical prediction model and an accompanying web tool to estimate the 3-month mortality risk for patients with anti-MDA5+DM-ILD.

Automated Fungal Identification with Deep Learning on Time-Lapse Images

The identification of species within filamentous fungi is crucial in various fields such as agriculture, environmental monitoring, and medical mycology. Traditional identification methods based on morphology have a low demand for advanced equipment usage and heavily depend on manual observation and expertise. However, this approach may struggle to differentiate between species in a genus due to their potential visual similarities, making the process time-consuming and subjective. In this study, we present an AI-based fungal species recognition model that utilizes deep learning techniques applied to time-lapse images. The training dataset, derived from fungi strains in the IBT Culture Collection, comprised 26,451 high-resolution images representing 110 species from 35 genera. The dataset was divided into a training set and validation subsets. We implemented three advanced deep learning architectures—ResNet50, DenseNet-121, and Vision Transformer (ViT)—to assess their effectiveness in accurately classifying fungal species. By utilizing images from early growth stages (days 2–3.5) for training and testing and later stages (days 4–7) for validation, our approach shortens the fungal identification process by 2–3 days, significantly reducing the associated workload and costs. Among the models, the Vision Transformer achieved the highest accuracy of 92.6%, demonstrating the effectiveness of our method. This work contributes to the automation of fungal identification, providing a reliable and efficient solution for monitoring fungal growth and diversity over time, which would be useful for culture collections or other institutions that handle a large number of new isolates in their daily work.

Feasibility and accuracy of single maxillary molar designed by an implicit neural network (INN)-based model: A comparative study.

This feasibility study aimed to develop and evaluate a novel implicit neural network (INN)-based model for the automatic reconstruction of biomimetic maxillary molar. A total of 500 sets of full dental scans containing intact right and left maxillary first molars (#3 and #14) and adjacent teeth were included in this study. The digital maxillary casts were duplicated: one set served as the original, while the other had one maxillary first molar removed. Two INN-based models were developed, including the point convolution for surface reconstruction (POCO-only) model and the POCO with point multilayer perceptron (POCO-PointMLP) model. Each model was trained with either 12,000 or 50,000 sampling points from the training dataset. Chamfer distance (CD), F-score, and Volumetric Intersection over Union (volumetric IoU) were used to evaluate the INN-based models and training outcomes. The best INN-based model was selected to generate 20 digital crown designs (GCs) automatically from randomly selected test datasets and compared with the digital crowns designed by experienced human laboratory technicians (TCs). Both GCs and TCs were compared with the original clinical crowns (OCs) using rootmean square(RMS) and color maps. Comparisons of RMS values were analyzed utilizing paired t-tests with statistical significance set at α=0.05. The performance of 2 INN-based models and training outcomes were evaluated and POCO-PointMLP trained with 50,000 sample points showed the best outcomes, with CD of 0.000119403, F-score of 0.919740566, and volumetric IoU of 0.943382978. This INN-based model was then selected to generate 20 digital crown designs (GCs) and compared with TCs for the RMS outcomes. The GCs exhibited numerically lower mean RMS compared to TCs without significant difference (0.2839±0.0307 versus 0.3026±0.0587mm, p=0.202). GCs and TCs both closely matched to the original clinical crowns (OCs). This study demonstrated the feasibility and accuracy of the proposed INN-based POCO-PointMLP model for automating the digital crown designs of maxillary first molars. The INN-based model-generated digital crown designs showed comparable 3D deviations to the ones designed by experienced human laboratory technicians, presenting a novel and efficient artificial neural network architecture for tooth digital design tasks with the expansion of training datasets, refined training hyperparameters, and integrated comprehensive assessments.

SPOCK 2.0: Updates to the FeatureClassifier in the Stability of Planetary Orbital Configurations Klassifier

Abstract The Stability of Planetary Orbital Configurations Klassifier (SPOCK) package collects machine learning models for predicting the stability and collisional evolution of compact planetary systems. In this paper we explore improvements to SPOCK’s binary stability classifier (FeatureClassifier), which predicts orbital stability by collecting data over a short N-body integration of a system. We find that by using a system-specific timescale (rather than a fixed 104 orbits) for the integration, and by using this timescale as an additional feature, we modestly improve the model’s AUC metric from 0.943 to 0.950 (AUC = 1 for a perfect model). We additionally discovered that ≈10% of N-body integrations in SPOCK’s original training data set were duplicated by accident, and that &lt;1% were misclassified as stable when they in fact led to ejections. We provide a cleaned data set of 100,000+ unique integrations, release a newly trained stability classification model, and make minor updates to the API.

PIPENN-EMB ensemble net and protein embeddings generalise protein interface prediction beyond homology

Protein interactions are crucial for understanding biological functions and disease mechanisms, but predicting these remains a complex task in computational biology. Increasingly, Deep Learning models are having success in interface prediction. This study presents PIPENN-EMB which explores the added value of using embeddings from the ProtT5-XL protein language model. Our results show substantial improvement over the previously published PIPENN model for protein interaction interface prediction, reaching an MCC of 0.313 vs. 0.249, and AUROC 0.800 vs. 0.755 on the BIO_DL_TE test set. We furthermore show that these embeddings cover a broad range of ‘hand-crafted’ protein features in ablation studies. PIPENN-EMB reaches state-of-the-art performance on the ZK448 dataset for protein-protein interface prediction. We showcase predictions on 25 resistance-related proteins from Mycobacterium tuberculosis. Furthermore, whereas other state-of-the-art sequence-based methods perform worse for proteins that have little recognisable homology in their training data, PIPENN-EMB generalises to remote homologs, yielding stable AUROC across all three test sets with less than 30% sequence identity to the training dataset, and even to proteins with less than 15% sequence identity.