Spectro-environmental factors integrated ensemble learning for urban river network water quality remote sensing.

Remote sensing water quality monitoring technology can effectively supplement the shortcomings of traditional water quality monitoring methods in spatiotemporal dynamic monitoring capabilities. At present, although the spectral feature-based remote sensing water quality inversion models have achieved many successes, there could still be a problem of insufficient generalization ability in monitoring the water quality of complex river networks in large cities. In this paper, we propose a spectro-environmental factors integrated ensemble learning model for urban river network water quality inversion. We analyzed the correlation between water quality parameters, spectral reflectance, and environmental factors based on an in-situ dataset collected in the northern part of Shanghai. Using the Hot Spot Analysis (Getis-Ord Gi*), we found that river network water quality parameters have different patterns in different urban functional zones. Furthermore, daily average temperature, total rainfall within the seven days, and several band combinations were also selected as the environmental and spectral features using factor analysis and Pearson correlation coefficient analysis. After the feature analysis, the spectro-environmental factors integrated ensemble learning model was trained. Compared with the spectral-based machine learning inversion models, the coefficients of determination R2 increased by about 0.50. Our model was also tested in three different test areas within and outside the in-situ sampling areas in Shanghai based on low-altitude multispectral remote sensing images. The R2 results for total phosphorus (TP), ammonia nitrogen (NH3-N), and chemical oxygen demand (COD) within the in-situ sampling areas were 0.52, 0.58, and 0.56 respectively. The mean absolute percentage error (MAPE) results were 53.36%, 63.95%, and 22.46% respectively. After adding the area outside the in-situ sampling areas, the R2 results for TP, NH3-N, and COD were 0.47, 0.47, and 0.53. The MAPE were 49.38%, 74.46%, and 20.49%. Our research provided a new remote sensing water quality inversion method to be utilized in complex urban river networks which exhibited solid accuracy and generalization ability.

Optimizing and benchmarking polygenic risk scores with GWAS summary statistics.

BACKGROUND: Polygenic risk score (PRS) is a major research topic in human genetics. However, a significant gap exists between PRS methodology and applications in practice due to often unavailable individual-level data for various PRS tasks including model fine-tuning, benchmarking, and ensemble learning. RESULTS: We introduce an innovative statistical framework to optimize and benchmark PRS models using summary statistics of genome-wide association studies. This framework builds upon our previous work and can fine-tune virtually all existing PRS models while accounting for linkage disequilibrium. In addition, we provide an ensemble learning strategy named PUMAS-ensemble to combine multiple PRS models into an ensemble score without requiring external data for model fitting. Through extensive simulations and analysis of many complex traits in the UK Biobank, we demonstrate that this approach closely approximates gold-standard analytical strategies based on external validation, and substantially outperforms state-of-the-art PRS methods. CONCLUSIONS: Our method is a powerful and general modeling technique that can continue to combine the best-performing PRS methods out there through ensemble learning and could become an integral component for all future PRS applications.

Topic selection for text classification using ensemble topic modeling with grouping, scoring, and modeling approach.

TextNetTopics (Yousef et al. in Front Genet 13:893378, 2022. https://doi.org/10.3389/fgene.2022.893378 ) is a recently developed approach that performs text classification-based topics (a topic is a group of terms or words) extracted from a Latent Dirichlet Allocation topic modeling as features rather than individual words. Following this approach enables TextNetTopics to fulfill dimensionality reduction while preserving and embedding more thematic and semantic information into the text document representations. In this article, we introduced a novel approach, the Ensemble Topic Model for Topic Selection (ENTM-TS), an advancement of TextNetTopics. ENTM-TS integrates multiple topic models using the Grouping, Scoring, and Modeling approach, thereby mitigating the performance variability introduced by employing individual topic modeling methods within TextNetTopics. Additionally, we performed a thorough comparative study to evaluate TextNetTopics' performance using eleven state-of-the-art topic modeling algorithms. We used the extracted topics for each as input to the G component in the TextNetTopics tool to select the most compelling topic model regarding their predictive behavior for text classification. We conducted our comprehensive evaluation utilizing the Drug-Induced Liver Injury textual dataset from the CAMDA community and the WOS-5736 dataset. The experimental results show that the Latent Semantic Indexing provides comparable performance measures with fewer discriminative features when compared with other topic modeling methods. Moreover, our evaluation reveals that the performance of ENTM-TS surpasses or aligns with the optimal outcomes obtained from individual topic models across the two datasets, establishing it as a robust and effective enhancement in text classification tasks.

Augmented histopathology: Enhancing colon cancer detection through deep learning and ensemble techniques.

Colon cancer poses a significant threat to human life with a high global mortality rate. Early and accurate detection is crucial for improving treatment quality and the survival rate. This paper presents a comprehensive approach to enhance colon cancer detection and classification. The histopathological images are gathered from the CRC-VAL-HE-7K dataset. The images undergo preprocessing to improve quality, followed by augmentation to increase dataset size and enhance model generalization. A deep learning based transformer model is designed for efficient feature extraction and enhancing classification by incorporating a convolutional neural network (CNN). A cross-transformation model captures long-range dependencies between regions, and an attention mechanism assigns weights to highlight crucial features. To boost classification accuracy, a Siamese network distinguishes colon cancer tissue classes based on probabilities. Optimization algorithms fine-tune model parameters, categorizing colon cancer tissues into different classes. The multi-class classification performance is evaluated in the experimental evaluation, which demonstrates that the proposed model provided highest accuracy rate of 98.84%. In this research article, the proposed method achieved better performance in all analyses by comparing with other existing methods. RESEARCH HIGHLIGHTS: Deep learning-based techniques are proposed. DL methods are used to enhance colon cancer detection and classification. CRC-VAL-HE-7K dataset is utilized to enhance image quality. Hybrid particle swarm optimization (PSO) and dwarf mongoose optimization (DMO) are used. The deep learning models are tuned by implementing the PSO-DMO algorithm.

SwAV-driven diagnostics: new perspectives on grading diabetic retinopathy from retinal photography.

Diabetic Retinopathy (DR) is a serious eye condition that occurs due to high blood sugar levels in patients with Diabetes Mellitus. If left untreated, DR can potentially result in blindness. Using automated neural network-based methods to grade DR shows potential for early detection. However, the uneven and non-quadrilateral forms of DR lesions provide difficulties for traditional Convolutional Neural Network (CNN)-based architectures. To address this challenge and explore a novel algorithm architecture, this work delves into the usage of contrasting cluster assignments in retinal fundus images with the Swapping Assignments between multiple Views (SwAV) algorithm for DR grading. An ablation study was made where SwAV outperformed other CNN and Transformer-based models, independently and in ensemble configurations with an accuracy of 87.00% despite having fewer parameters and layers. The proposed approach outperforms existing state-of-the-art models regarding classification metrics, complexity, and prediction time. The findings offer great potential for medical practitioners, allowing for more accurate diagnosis of DR and earlier treatments to avoid visual loss.

Dynamic stacking ensemble for cross-language code smell detection.

Code smells refer to poor design and implementation choices by software engineers that might affect the overall software quality. Code smells detection using machine learning models has become a popular area to build effective models that are capable of detecting different code smells in multiple programming languages. However, the process of building of such effective models has not reached a state of stability, and most of the existing research focuses on Java code smells detection. The main objective of this article is to propose dynamic ensembles using two strategies, namely greedy search and backward elimination, which are capable of accurately detecting code smells in two programming languages (i.e., Java and Python), and which are less complex than full stacking ensembles. The detection performance of dynamic ensembles were investigated within the context of four Java and two Python code smells. The greedy search and backward elimination strategies yielded different base models lists to build dynamic ensembles. In comparison to full stacking ensembles, dynamic ensembles yielded less complex models when they were used to detect most of the investigated Java and Python code smells, with the backward elimination strategy resulting in less complex models. Dynamic ensembles were able to perform comparably against full stacking ensembles with no significant detection loss. This article concludes that dynamic stacking ensembles were able to facilitate the effective and stable detection performance of Java and Python code smells over all base models and with less complexity than full stacking ensembles.

Enhancing intrusion detection performance using explainable ensemble deep learning.

Given the exponential growth of available data in large networks, the need for an accurate and explainable intrusion detection system has become of high necessity to effectively discover attacks in such networks. To deal with this challenge, we propose a two-phase Explainable Ensemble deep learning-based method (EED) for intrusion detection. In the first phase, a new ensemble intrusion detection model using three one-dimensional long short-term memory networks (LSTM) is designed for an accurate attack identification. The outputs of three classifiers are aggregated using a meta-learner algorithm resulting in refined and improved results. In the second phase, interpretability and explainability of EED outputs are enhanced by leveraging the capabilities of SHape Additive exPplanations (SHAP). Factors contributing to the identification and classification of attacks are highlighted which allows security experts to understand and interpret the attack behavior and then implement effective response strategies to improve the network security. Experiments conducted on real datasets have shown the effectiveness of EED compared to conventional intrusion detection methods in terms of both accuracy and explainability. The EED method exhibits high accuracy in accurately identifying and classifying attacks while providing transparency and interpretability.

An ensemble learning model for predicting cancer-specific survival of muscle-invasive bladder cancer patients undergoing bladder preservation therapy.

Background: More muscle-invasive bladder cancer (MIBC) patients are now eligible for bladder-preserving therapy (BPT), underscoring the need for precision medicine. This study aimed to identify prognostic predictors and construct a predictive model among MIBC patients who undergo BPT. Methods: Data relating to MIBC patients were obtained from the Surveillance, Epidemiology and End Results (SEER) database from 2004 to 2016. Eleven features were included to establish multiple models. The predictive effectiveness was assessed using receiver operating characteristic (ROC) curves, calibration plots, decision curve analysis (DCA) and clinical impact curve (CIC). SHapley Additive exPlanations (SHAP) were used to explain the impact of features on the predicted targets. Results: The ROC showed that Catboost and Random Forest (RF) obtained better predictive discrimination in both 3- and 5-year models [test set area under curves (AUC) =0.80 and 0.83, respectively]. Furthermore, Catboost showed better performance in calibration plots, DCA and CIC. SHAP analysis indicated that age, M stage, tumor size, chemotherapy, T stage and gender were the most important features in the model for predicting the 3-year cancer-specific survival (CSS). In contrast, M stage, age, tumor size and gender as well as the N and T stages were the most important features for predicting the 5-year CSS. Conclusions: The Catboost model exhibits the highest predictive performance and clinical utility, potentially aiding clinicians in making optimal individualized decisions for MIBC patients with BPT.

An ensemble approach integrating LSTM and ARIMA models for enhanced financial market predictions.

Forecasting financial markets is a complex task that requires addressing various challenges, such as market complexity, data heterogeneity, the need for rapid response and constant changes in conditions, to gain a competitive advantage. To effectively address these challenges, it is necessary to constantly improve existing and develop new methods of intelligent forecasting, which will improve the accuracy of forecasts, reduce risks and increase the productivity of financial decision-making processes. In this article, we study and analyse forecasting methods in financial markets, such as support vector regression (SVR), autoregressive integrated moving average (ARIMA), long short-term memory recurrent neural network (LSTM) and extreme gradient boosting algorithm (XG-Boost). Based on this analysis, we propose an ensemble forecasting procedure that integrates LSTM and ARIMA models. Due to the careful combination of these models, our approach yields better results than individual methods. For example, our model demonstrates a significant 15% improvement in root mean square error (RMSE) and a slight improvement in coefficient of determination compared with LSTM. Furthermore, simulation results obtained on three real-world datasets and evaluated using the RMSE criterion confirm the superiority of our proposed method over alternative methods such as LSTMs, transformer models and optimized deep recurrent neural networks with long short-term memory for financial market forecasting. Furthermore, our approach creates the prerequisites for parallelizing both models, thus providing an opportunity to accelerate forecasting results in future research.

Noise-induced modality-specific pretext learning for pediatric chest X-ray image classification.

Introduction: Deep learning (DL) has significantly advanced medical image classification. However, it often relies on transfer learning (TL) from models pretrained on large, generic non-medical image datasets like ImageNet. Conversely, medical images possess unique visual characteristics that such general models may not adequately capture. Methods: This study examines the effectiveness of modality-specific pretext learning strengthened by image denoising and deblurring in enhancing the classification of pediatric chest X-ray (CXR) images into those exhibiting no findings, i.e., normal lungs, or with cardiopulmonary disease manifestations. Specifically, we use a VGG-16-Sharp-U-Net architecture and leverage its encoder in conjunction with a classification head to distinguish normal from abnormal pediatric CXR findings. We benchmark this performance against the traditional TL approach, viz., the VGG-16 model pretrained only on ImageNet. Measures used for performance evaluation are balanced accuracy, sensitivity, specificity, F-score, Matthew's Correlation Coefficient (MCC), Kappa statistic, and Youden's index. Results: Our findings reveal that models developed from CXR modality-specific pretext encoders substantially outperform the ImageNet-only pretrained model, viz., Baseline, and achieve significantly higher sensitivity (p < 0.05) with marked improvements in balanced accuracy, F-score, MCC, Kappa statistic, and Youden's index. A novel attention-based fuzzy ensemble of the pretext-learned models further improves performance across these metrics (Balanced accuracy: 0.6376; Sensitivity: 0.4991; F-score: 0.5102; MCC: 0.2783; Kappa: 0.2782, and Youden's index:0.2751), compared to Baseline (Balanced accuracy: 0.5654; Sensitivity: 0.1983; F-score: 0.2977; MCC: 0.1998; Kappa: 0.1599, and Youden's index:0.1327). Discussion: The superior results of CXR modality-specific pretext learning and their ensemble underscore its potential as a viable alternative to conventional ImageNet pretraining for medical image classification. Results from this study promote further exploration of medical modality-specific TL techniques in the development of DL models for various medical imaging applications.

Fluorescence hyperspectral imaging technology combined with chemometrics for kiwifruit quality attribute assessment and non-destructive judgment of maturity.

Dry matter content (DMC), firmness and soluble solid content (SSC) are important indicators for assessing the quality attributes and determining the maturity of kiwifruit. However, traditional measurement methods are time-consuming, labor-intensive, and destructive to the kiwifruit, leading to resource wastage. In order to solve this problem, this study has tracked the flowering, fruiting, maturing and collecting processes of Ya'an red-heart kiwifruit, and has proposed a non-destructive method for kiwifruit quality attribute assessment and maturity identification that combines fluorescence hyperspectral imaging (FHSI) technology and chemometrics. Specifically, first of all, three different spectral data preprocessing methods were adopted, and PLSR was used to evaluate the quality attributes (DMC, firmness, and SSC) of kiwifruit. Next, the differences in accuracy of different models in discriminating kiwifruit maturity were compared, and an ensemble learning model based on LightGBM and GBDT models was constructed. The results indicate that the ensemble learning model outperforms single machine learning models. In addition, the application effects of the 'Convolutional Neural Network'-'Multilayer Perceptron' (CNN-MLP) model under different optimization algorithms were compared. To improve the robustness of the model, an improved whale optimization algorithm (IWOA) was introduced by modifying the acceleration factor. Overall, the IWOA-CNN-MLP model performs the best in discriminating the maturity of kiwifruit, with Accuracytest of 0.916 and Loss of 0.23. In addition, compared with the basic model, the accuracy of the integrated learning model SG-MSC-SEL was improved by about 12%-20 %. The research findings will provide new perspectives for the evaluation of kiwifruit quality and maturity discrimination using FHSI and chemometric methods, thereby promoting further research and applications in this field.

RV-FELM: Futures commodity price forecasting based on RIME-VMD algorithm coupled with FA-ELM.

Commodity futures are an important hedging tool in material trade, and by accurately predicting prices, countries and firms are able to make informed production and consumption decisions. This paper introduces a novel machine learning ensemble method that combines decomposition algorithms and physical optimization algorithms to predict commodity futures prices. First, the VMD(Variational mode decomposition) is optimized by the RIME algorithm (Rime optimization algorithm) to obtain the optimal modal decomposition results, and the trend and seasonal terms are predicted using the ELM (Extreme Learning Machines) and FA (Fourier Attention) models, respectively, and the results are finally synthesized. The results show that the MAPE(mean absolute percentage error) of one-step, three-step, and six-step methods for predicting crude oil prices are 0.48%, 0.66%, and 0.75%, respectively, and the MAPE of soybean prediction results are 0.22%, 0.27%, and 0.37%, respectively. The empirical results and ablation experiments show that it outperforms other benchmark models in terms of both horizontal and directional accuracy. Notably, it outperforms in predicting soybean futures prices, which demonstrates the ability of our model to better capture the characteristics of both the time and frequency domains of the series, to take sufficient consideration of the series characteristics, and to ensure robustness.

Domain-knowledge enabled ensemble learning of 5-formylcytosine (f5C) modification sites.

5-formylcytidine (f5C) is a unique post-transcriptional RNA modification found in mRNA and tRNA at the wobble site, playing a crucial role in mitochondrial protein synthesis and potentially contributing to the regulation of translation. Recent studies have unveiled that the f5C modifications may drive mitochondrial mRNA translation to power cancer metastasis. Accurate identification of f5C sites is essential for further unraveling their molecular functions and regulatory mechanisms, but there are currently no computational methods available for predicting their locations. In this study, we introduce an innovative ensemble approach, successfully enabling the computational recognition of Saccharomyces cerevisiae f5C. We conducted a comprehensive model selection process that involved multiple basic machine learning and deep learning algorithms such as recurrent neural networks, convolutional neural networks and Transformer-based models. Initially trained only on sequence information, these individual models achieved an AUROC ranging from 0.7104 to 0.7492. Through the integration of 32 novel domain-derived genomic features, the performance of individual models has significantly improved to an AUROC between 0.7309 and 0.8076. To further enhance accuracy and robustness, we then constructed the ensembles of these individual models with different combinations. The best performance attained by our ensemble models reached an AUROC of 0.8391. Shapley additive explanations were conducted to explain the significant contributions of genomic features, providing insights into the putative distribution of f5C across various topological regions and potentially paving the way for revealing their functional relevance within distinct genomic contexts. A freely accessible web server that allows real-time analysis of user-uploaded sites can be accessed at: www.rnamd.org/Resf5C-Pred.

Adaptive 3DCNN-based Interpretable Ensemble Model for Early Diagnosis of Alzheimer's Disease.

Adaptive interpretable ensemble model based on three-dimensional Convolutional Neural Network (3DCNN) and Genetic Algorithm (GA), i.e., 3DCNN+EL+GA, was proposed to differentiate the subjects with Alzheimer's Disease (AD) or Mild Cognitive Impairment (MCI) and further identify the discriminative brain regions significantly contributing to the classifications in a data-driven way. Plus, the discriminative brain sub-regions at a voxel level were further located in these achieved brain regions, with a gradient-based attribution method designed for CNN. Besides disclosing the discriminative brain sub-regions, the testing results on the datasets from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and the Open Access Series of Imaging Studies (OASIS) indicated that 3DCNN+EL+GA outperformed other state-of-the-art deep learning algorithms and that the achieved discriminative brain regions (e.g., the rostral hippocampus, caudal hippocampus, and medial amygdala) were linked to emotion, memory, language, and other essential brain functions impaired early in the AD process. Future research is needed to examine the generalizability of the proposed method and ideas to discern discriminative brain regions for other brain disorders, such as severe depression, schizophrenia, autism, and cerebrovascular diseases, using neuroimaging.

Dynamic weighted residual ensemble learning for hyperspectral image classification driven by features and samples.

Dynamic ensemble selection has emerged as a promising approach for hyperspectral image classification. However, selecting relevant features and informative samples remains a pressing challenge. To address this issue, we introduce two novel dynamic residual ensemble learning methods. The first proposed method is called multi-features driven dynamic weighted residuals ensemble learning (MF-DWRL). This method leverages various combinations of features to construct classifier pools that incorporate feature differences. The K-Nearest Neighbors algorithm is employed to establish the region of competence (RoC) in the dynamic ensemble selection process. By assessing the performance of the RoC, the feature sets that yield the highest classification accuracy are identified as the optimal feature combinations. Additionally, the classification accuracy is utilized as prior information to guide the residual adjustments of each classifier. The second method, known as features and samples double-driven dynamic weighted residual ensemble learning (FS-DWRL), further enhances the performance of the ensemble. This approach not only considers the selection of feature combinations but also takes into account the informative samples. By jointly optimizing the feature and sample selection processes, FS-DWRL achieves superior classification accuracy compared to existing state-of-the-art methods. To evaluate the effectiveness of the proposed methods, three hyperspectral datasets from China-WHU-Hi-HanChuan, WHU-Hi-LongKou, and WHU-Hi-HongHu-are used for classification experiments. For these datasets, the proposed methods achieve the highest classification accuracies of 90.57 %, 98.77 %, and 91.08 %, respectively. The MF-DWRL and FS-DWRL methods exhibit significant improvements in classification accuracy.

Integrating Laser-Induced Breakdown Spectroscopy and Ensemble Learning as Minimally Invasive Optical Screening for Diabetes.

Diabetes mellitus is a prevalent chronic disease necessitating timely identification for effective management. This paper introduces a reliable, straightforward, and efficient method for the minimally invasive identification of diabetes mellitus through nanosecond pulsed laser-induced breakdown spectroscopy (LIBS) by integrating a state-of-the-art machine learning approach. LIBS spectra were collected from urine samples of diabetic and healthy individuals. Principal component analysis and an ensemble learning classification model were used to identify significant changes in LIBS peak intensity between the diseased and normal urine samples. The model, integrating six distinct classifiers and cross-validation techniques, exhibited high accuracy (96.5%) in predicting diabetes mellitus. Our findings emphasize the potential of LIBS for diabetes mellitus identification in urine samples. This technique may hold potential for future applications in diagnosing other health conditions.

Machine-learning-based prediction by stacking ensemble strategy for surgical outcomes in patients with degenerative cervical myelopathy.

BACKGROUND: Machine learning (ML) is extensively employed for forecasting the outcome of various illnesses. The objective of the study was to develop ML based classifiers using a stacking ensemble strategy to predict the Japanese Orthopedic Association (JOA) recovery rate for patients with degenerative cervical myelopathy (DCM). METHODS: A total of 672 patients with DCM were included in the study and labeled with JOA recovery rate by 1-year follow-up. All data were collected during 2012-2023 and were randomly divided into training and testing (8:2) sub-datasets. A total of 91 initial ML classifiers were developed, and the top 3 initial classifiers with the best performance were further stacked into an ensemble classifier with a supported vector machine (SVM) classifier. The area under the curve (AUC) was the main indicator to assess the prediction performance of all classifiers. The primary predicted outcome was the JOA recovery rate. RESULTS: By applying an ensemble learning strategy (e.g., stacking), the accuracy of the ML classifier improved following combining three widely used ML models (e.g., RFE-SVM, EmbeddingLR-LR, and RFE-AdaBoost). Decision curve analysis showed the merits of the ensemble classifiers, as the curves of the top 3 initial classifiers varied a lot in predicting JOA recovery rate in DCM patients. CONCLUSIONS: The ensemble classifiers successfully predict the JOA recovery rate in DCM patients, which showed a high potential for assisting physicians in managing DCM patients and making full use of medical resources.

Early Prediction of Cardiac Arrest in the Intensive Care Unit Using Explainable Machine Learning: Retrospective Study.

BACKGROUND: Cardiac arrest (CA) is one of the leading causes of death among patients in the intensive care unit (ICU). Although many CA prediction models with high sensitivity have been developed to anticipate CA, their practical application has been challenging due to a lack of generalization and validation. Additionally, the heterogeneity among patients in different ICU subtypes has not been adequately addressed. OBJECTIVE: This study aims to propose a clinically interpretable ensemble approach for the timely and accurate prediction of CA within 24 hours, regardless of patient heterogeneity, including variations across different populations and ICU subtypes. Additionally, we conducted patient-independent evaluations to emphasize the model's generalization performance and analyzed interpretable results that can be readily adopted by clinicians in real-time. METHODS: Patients were retrospectively analyzed using data from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) and the eICU-Collaborative Research Database (eICU-CRD). To address the problem of underperformance, we constructed our framework using feature sets based on vital signs, multiresolution statistical analysis, and the Gini index, with a 12-hour window to capture the unique characteristics of CA. We extracted 3 types of features from each database to compare the performance of CA prediction between high-risk patient groups from MIMIC-IV and patients without CA from eICU-CRD. After feature extraction, we developed a tabular network (TabNet) model using feature screening with cost-sensitive learning. To assess real-time CA prediction performance, we used 10-fold leave-one-patient-out cross-validation and a cross-data set method. We evaluated MIMIC-IV and eICU-CRD across different cohort populations and subtypes of ICU within each database. Finally, external validation using the eICU-CRD and MIMIC-IV databases was conducted to assess the model's generalization ability. The decision mask of the proposed method was used to capture the interpretability of the model. RESULTS: The proposed method outperformed conventional approaches across different cohort populations in both MIMIC-IV and eICU-CRD. Additionally, it achieved higher accuracy than baseline models for various ICU subtypes within both databases. The interpretable prediction results can enhance clinicians' understanding of CA prediction by serving as a statistical comparison between non-CA and CA groups. Next, we tested the eICU-CRD and MIMIC-IV data sets using models trained on MIMIC-IV and eICU-CRD, respectively, to evaluate generalization ability. The results demonstrated superior performance compared with baseline models. CONCLUSIONS: Our novel framework for learning unique features provides stable predictive power across different ICU environments. Most of the interpretable global information reveals statistical differences between CA and non-CA groups, demonstrating its utility as an indicator for clinical decisions. Consequently, the proposed CA prediction system is a clinically validated algorithm that enables clinicians to intervene early based on CA prediction information and can be applied to clinical trials in digital health.

A novel deep ensemble-based model with outlier removal and order-invariant ranking for carbon dioxide emission prediction.

Excessive carbon dioxide ( CO 2 ) emissions pose a formidable challenge, driving global climate change and necessitating urgent attention. Striking a balance between curbing CO 2 emissions and fostering economic growth hinges upon the ability to reliably forecast CO 2 emissions. Such forecasts are indispensable for policymakers as they endeavor to make informed decisions and proactively implement mitigation measures. In this research, we introduce an innovative deep ensemble prediction model for CO 2 emissions. This model is constructed around four parallel Long Short-Term Memory (LSTM) neural networks, complemented by a novel Multi-Layer Perception (MLP)-based ensemble framework, equipped with an outlier detection mechanism and an order-invariant ranking module. To enhance prediction accuracy and stability, a k-nearest neighbor (KNN)-based outlier detection module is employed to identify non-outliers and reasonable predictions for the ensemble models. Additionally, a novel feature ranking module is proposed to mitigate prediction fluctuations. The performance evaluation of our model is conducted using historical CO 2 emission data spanning from 1971 to 2021, encompassing six representative countries. Our findings demonstrate that the proposed methodology outperforms existing approaches across various evaluation metrics, offering considerably reduced prediction variances and greater stability. Moreover, long-term CO 2 emission predictions for the corresponding six countries have been provided, which might offer policymakers some basis for making decisions.

A deep-learning pipeline for the diagnosis and grading of common blinding ophthalmic diseases based on lesion-focused classification model.

Background: Glaucoma (GLAU), Age-related Macular Degeneration (AMD), Retinal Vein Occlusion (RVO), and Diabetic Retinopathy (DR) are common blinding ophthalmic diseases worldwide. Purpose: This approach is expected to enhance the early detection and treatment of common blinding ophthalmic diseases, contributing to the reduction of individual and economic burdens associated with these conditions. Methods: We propose an effective deep-learning pipeline that combine both segmentation model and classification model for diagnosis and grading of four common blinding ophthalmic diseases and normal retinal fundus. Results: In total, 102,786 fundus images of 75,682 individuals were used for training validation and external validation purposes. We test our model on internal validation data set, the micro Area Under the Receiver Operating Characteristic curve (AUROC) of which reached 0.995. Then, we fine-tuned the diagnosis model to classify each of the four disease into early and late stage, respectively, which achieved AUROCs of 0.597 (GL), 0.877 (AMD), 0.972 (RVO), and 0.961 (DR) respectively. To test the generalization of our model, we conducted two external validation experiments on Neimeng and Guangxi cohort, all of which maintained high accuracy. Conclusion: Our algorithm demonstrates accurate artificial intelligence diagnosis pipeline for common blinding ophthalmic diseases based on Lesion-Focused fundus that overcomes the low-accuracy of the traditional classification method that based on raw retinal images, which has good generalization ability on diverse cases in different regions.