Performance Of Model Research Articles

Reproducible Radiomics Features from Multi-MRI-Scanner Test-Retest-Study: Influence on Performance and Generalizability of Models.

Radiomics models trained on data from one center typically show a decline of performance when applied to data from external centers, hindering their introduction into large-scale clinical practice. Current expert recommendations suggest to use only reproducible radiomics features isolated by multiscanner test-retest experiments, which might help to overcome the problem of limited generalizability to external data. To evaluate the influence of using only a subset of robust radiomics features, defined in a prior in vivo multi-MRI-scanner test-retest-study, on the performance and generalizability of radiomics models. Retrospective. Patients with monoclonal plasma cell disorders. Training set (117 MRIs from center 1); internal test set (42 MRIs from center 1); external test set (143 MRIs from center 2-8). 1.5T and 3.0T; T1-weighted turbo spin echo. The task for the radiomics models was to predict plasma cell infiltration, determined by bone marrow biopsy, noninvasively from MRI. Radiomics machine learning models, including linear regressor, support vector regressor (SVR), and random forest regressor (RFR), were trained on data from center 1, using either all radiomics features, or using only reproducible radiomics features. Models were tested on an internal (center 1) and a multicentric external data set (center 2-8). Pearson correlation coefficient r and mean absolute error (MAE) between predicted and actual plasma cell infiltration. Fisher's z-transformation, Wilcoxon signed-rank test, Wilcoxon rank-sum test; significance level P < 0.05. When using only reproducible features compared with all features, the performance of the SVR on the external test set significantly improved (r = 0.43 vs. r = 0.18 and MAE = 22.6 vs. MAE = 28.2). For the RFR, the performance on the external test set deteriorated when using only reproducible instead of all radiomics features (r = 0.33 vs. r = 0.44, P = 0.29 and MAE = 21.9 vs. MAE = 20.5, P = 0.10). Using only reproducible radiomics features improves the external performance of some, but not all machine learning models, and did not automatically lead to an improvement of the external performance of the overall best radiomics model. Stage 2.

Sampling-Based Machine Learning Models for Intrusion Detection in Imbalanced Dataset

Cybersecurity is one of the important considerations when adopting IoT devices in smart applications. Even though a huge volume of data is available, data related to attacks are generally in a significantly smaller proportion. Although machine learning models have been successfully applied for detecting security attacks on smart applications, their performance is affected by the problem of such data imbalance. In this case, the prediction model is preferable to the majority class, while the performance for predicting the minority class is poor. To address such problems, we apply two oversampling techniques and two undersampling techniques to balance the data in different categories. To verify their performance, five machine learning models, namely the decision tree, multi-layer perception, random forest, XGBoost, and CatBoost, are used in the experiments based on the grid search with 10-fold cross-validation for parameter tuning. The results show that both the oversampling and undersampling techniques can improve the performance of the prediction models used. Based on the results, the XGBoost model based on the SMOTE has the best performance in terms of accuracy at 75%, weighted average precision at 82%, weighted average recall at 75%, weighted average F1 score at 78%, and Matthews correlation coefficient at 72%. This indicates that this oversampling technique is effective for multi-attack prediction under a data imbalance scenario.

Novel Learning of Bathymetry from Landsat 9 Imagery Using Machine Learning, Feature Extraction and Meta-Heuristic Optimization in a Shallow Turbid Lagoon

Bathymetry data is indispensable for a variety of aquatic field studies and benthic resource inventories. Determining water depth can be accomplished through an echo sounding system or remote estimation utilizing space-borne and air-borne data across diverse environments, such as lakes, rivers, seas, or lagoons. Despite being a common option for bathymetry mapping, the use of satellite imagery faces challenges due to the complex inherent optical properties of water bodies (e.g., turbid water), satellite spatial resolution limitations, and constraints in the performance of retrieval models. This study focuses on advancing the remote sensing based method by harnessing the non-linear learning capabilities of the machine learning (ML) model, employing advanced feature selection through a meta-heuristic algorithm, and using image extraction techniques (i.e., band ratio, gray scale morphological operation, and morphological multi-scale decomposition). Herein, we validate the predictive capabilities of six ML models: Random Forest (RF), Support Vector Machine (SVM), CatBoost (CB), Extreme Gradient Boost (XGB), Light Gradient Boosting Machine (LGBM), and KTBoost (KTB) models, both with and without the application of meta-heuristic optimization (i.e., Dragon Fly, Particle Swarm Optimization, and Grey Wolf Optimization), to accurately ascertain water depth. This is achieved using a diverse input dataset derived from multi-spectral Landsat 9 imagery captured on a cloud-free day (19 September 2023) in a shallow, turbid lagoon. Our findings indicate the superior performance of LGBM coupled with Particle Swamp Optimization (R2 = 0.908, RMSE = 0.31 m), affirming the consistency and reliability of the feature extraction and selection-based framework, while offering novel insights into the expansion of bathymetric mapping in complex aquatic environments.

TinyNS: Platform-aware Neurosymbolic Auto Tiny Machine Learning

Machine learning at the extreme edge has enabled a plethora of intelligent, time-critical, and remote applications. However, deploying interpretable artificial intelligence systems that can perform high-level symbolic reasoning and satisfy the underlying system rules and physics within the tight platform resource constraints is challenging. In this article, we introduce TinyNS , the first platform-aware neurosymbolic architecture search framework for joint optimization of symbolic and neural operators. TinyNS provides recipes and parsers to automatically write microcontroller code for five types of neurosymbolic models, combining the context awareness and integrity of symbolic techniques with the robustness and performance of machine learning models. TinyNS uses a fast, gradient-free, black-box Bayesian optimizer over discontinuous, conditional, numeric, and categorical search spaces to find the best synergy of symbolic code and neural networks within the hardware resource budget. To guarantee deployability, TinyNS talks to the target hardware during the optimization process. We showcase the utility of TinyNS by deploying microcontroller-class neurosymbolic models through several case studies. In all use cases, TinyNS outperforms purely neural or purely symbolic approaches while guaranteeing execution on real hardware.

Performance evaluation of varies climate models using observed and regional climate models for the Katar Watershed, Ethiopia

Climate models are fundamental tools to estimates the reliable future climate change and its effects on the water resources and agriculture in basins. However, all climate models are not equally performed for all areas. Therefore, determining the most appropriate climate models for a specific study area is essential. The focus of this study was to evaluate the performance of the regional climate models with regard to simulating precipitation, and temperatures at Katar watershed. This study examines the performance of fourteen CORDEX-AFRICA-220 Regional Climate Models (RCMs) for the period of 1984–2005 using statistical metrics such as Pearson correlation coefficient (R), mean absolute error (MAE), root mean squared error (RMSE), and bias. The findings indicated that GERICS-MPI was better performed in representing Areta, and Bokoji station, GERICS-IPSL was better representing in Assela, Ketergenet, and Sagure station, CCCma-CanESM2-AFR22, and RCA4-ICHEC performed relatively better in representing the mean annual observed rainfall at the Kulumsa, and Ogolcho station respectively. However, RCA4-CSIRO performed weakly in estimation of annual rainfall at all stations. RCM model such as GERICS-MPI was relatively better than the others in replicating the annual pattern of the maximum temperature at Areta, Bokoji, and Ketergenet stations. Similarly, GERICS-IPSL were relatively better in replicating the annual maximum temperature at Assela, and Sagure stations, CCCma-CanESM2-AFR22 at Kulumsa station, and RCA4-ICHEC at Ogolcho station performed well in capturing the observed and simulated annual maximum temperature. Better performance was observed on minimum temperature at CCCma-CanESM2-AFR22 at Areta, Assela, and Ketergenet stations, GERICS-MOHE-AFR-22 at Bokoji station, GERICS-MPI at Kulumsa, and Ogolcho stations, RAC4-NOAA-2G at Sagure stations. However, weak performance was observed RCA4-CSIRO at all stations. RCM models of GERICS-MPI, and CCLM4-NCC-AFR-22 performed better than the other RCM models for correction of annual rainfall in Katar watershed. However, poor performance was observed at RCA4-ICHEC model on Katar watershed. The GERICS-MPI model performed well. However, poor performance was observed at RCA4-ICHEC on maximum temperature, and GERICS-NOAA-2M on minimum temperature in Katar watershed.

Thermodiffusively unstable laminar hydrogen flame in a sufficiently large 3D computational domain – Part II: NOx formation mechanism and flamelet modeling

In this work, the NOx formation mechanism in three-dimensional (3D) thermodiffusively unstable premixed hydrogen flames is investigated through direct numerical simulations (DNS) and a reaction path analysis based on the DNS dataset. In part I of this study (Wen et al., 2024), the characteristic patterns of the instabilities were studied using the same dataset. Here, first, the effects of the computational setup (2D vs. 3D) and curvature on the flux ratio of nitrogen atom are quantified. In addition, the composition space model (CSM) proposed in previous works is extended to account for the NOx chemical reaction mechanism to predict NOx formation in thermodiffusively unstable premixed hydrogen flames. The flamelet solutions are obtained from the premixed flamelet equations in composition space so that the wide range of curvatures associated with the strongly corrugated flame front can be considered. The performance of the CSM in predicting the NOx species and the important radicals involved in the NOx formation pathways is evaluated through an a priori analysis. To investigate the effects of curvature on the NOx formation pathways and the performance of the composition space model, the positively- and negatively-curved regions in the DNS are studied separately. The results show that different from the 2D simulation, the NNH reaction pathway becomes dominant in the 3D simulation due to the increased range of curvature, which promotes the accumulation of the highly diffusive H radical in the positively-curved regions. The NNH reaction pathway is dominant in the positively-curved regions, while the N2O reaction pathway is more important in the negatively-curved regions. The flamelet model based on the composition space solutions that consider the effects of curvature yields accurate predictions for the radicals that are sensitive to the effects of curvature.Novelty and significance statementThe novelty of the present work includes the following aspects,(i) Analysis of NOx formation mechanism based on the large-scale 3D DNS dataset of a laminar fuel-lean premixed hydrogen flame on a sufficiently large domain;(ii) Quantification of the effects of computational setup (2D and 3D) and curvature on the NOx formation pathways;(iii) Extension of the previous composition space model (CSM) by incorporating the NOx chemical reaction mechanism to consider the effects of curvature on the NOx formation.

GPT-4 Turbo with Vision fails to outperform text-only GPT-4 Turbo in the Japan Diagnostic Radiology Board Examination.

To assess the performance of GPT-4 Turbo with Vision (GPT-4TV), OpenAI's latest multimodal large language model, by comparing its ability to process both text and image inputs with that of the text-only GPT-4 Turbo (GPT-4T) in the context of the Japan Diagnostic Radiology Board Examination (JDRBE). The dataset comprised questions from JDRBE 2021 and 2023. A total of six board-certified diagnostic radiologists discussed the questions and provided ground-truth answers by consulting relevant literature as necessary. The following questions were excluded: those lacking associated images, those with no unanimous agreement on answers, and those including images rejected by the OpenAI application programming interface. The inputs for GPT-4TV included both text and images, whereas those for GPT-4T were entirely text. Both models were deployed on the dataset, and their performance was compared using McNemar's exact test. The radiological credibility of the responses was assessed by two diagnostic radiologists through the assignment of legitimacy scores on a five-point Likert scale. These scores were subsequently used to compare model performance using Wilcoxon's signed-rank test. The dataset comprised 139 questions. GPT-4TV correctly answered 62 questions (45%), whereas GPT-4T correctly answered 57 questions (41%). A statistical analysis found no significant performance difference between the two models (P = 0.44). The GPT-4TV responses received significantly lower legitimacy scores from both radiologists than the GPT-4T responses. No significant enhancement in accuracy was observed when using GPT-4TV with image input compared with that of using text-only GPT-4T for JDRBE questions.

The Effects of Climate Change on Streamflow, Nitrogen Loads, and Crop Yields in the Gordes Dam Basin, Turkey

The Mediterranean region is highly vulnerable to climate change. Longer and more intense heatwaves and droughts are expected. The Gordes Dam in Turkey provides drinking water for Izmir city and irrigation water for a wide range of crops grown in the basin. Using the Soil and Water Assessment Tool (SWAT), this study examined the effects of projected climate change (RCP 4.5 and RCP 8.5) on the simulated streamflow, nitrogen loads, and crop yields in the basin for the period of 2031–2060. A hierarchical approach to define the hydrological response units (HRUs) of SWAT and the Fast Automatic Calibration Tool (FACT) were used to reduce computational time and improve model performance. The simulations showed that the average annual discharge into the reservoir is projected to increase by between 0.7 m3/s and 4 m3/s under RCP 4.5 and RCP 8.5 climate change scenarios. The steep slopes and changes in precipitation in the study area may lead to higher simulated streamflow. In addition, the rising temperatures predicted in the projections could lead to earlier spring snowmelt. This could also lead to increased streamflow. Projected nitrogen loads increased by between 8.8 and 25.1 t/year. The results for agricultural production were more variable. While the yields of poppy, tobacco, winter barley, and winter wheat will increase to some extent because of climate change, the yields of maize, cucumbers, and potatoes are all predicted to be negatively affected. Non-continuous and limited data on water quality and crop yields lead to uncertainties, so that the accuracy of the model is affected by these limitations and inconsistencies. However, the results of this study provide a basis for developing sustainable water and land management practices at the catchment scale in response to climate change. The changes in water quality and quantity and the ecological balance resulting from changes in land use and management patterns for economic benefit could not be fully demonstrated in this study. To explore the most appropriate management strategies for sustainable crop production, the SWAT model developed in this study should be further used in a multi-criteria land use optimization analysis that considers not only crop yields but also water quantity and quality targets.

Experimental and simulated evaluation of inverse model for shallow underground thermal storage

This study employs an inverse grey-box (IGB) modeling approach, which combines measured data and the physics of systems to predict the performance of shallow underground thermal energy storage (UTES) with top and side insulation. A simplified IGB model of the shallow UTES was developed using thermal network analysis. Experimental studies were conducted for shallow vertical and horizontal UTES configurations. Detailed models representing the field experiments were developed using the TRNSYS simulation tool and calibrated with the measured data from the experimental setup. The 4 Resistance 2 Capacitance (4R2C) IGB model was trained and tested in MATLAB using field experimental data and the data from the calibrated TRNSYS model. In comparing experimental data with the TRNSYS model for UTES, the prediction of outlet water temperature showed good agreement, with root mean square error (RMSE) and coefficient of variation of RMSE (CVRMSE) values of 0.94 °C and 3.16 % for vertical, and 0.99 °C and 2.97 % for horizontal configurations. The IGB model also aligned well with the experimental data, showing CVRMSE of 7.91 % and 3.17 % for vertical and horizontal systems, respectively. Sensitivity analysis revealed that model performance improves with longer training durations and closer testing times to training periods. However, a convergence point of 20 weeks of training data was achieved for making long-term performance predictions.

Systematic Review of Prognosis Models in Predicting Tooth Loss in Periodontitis.

This study reviews and appraises the methodological and reporting quality of prediction models for tooth loss in periodontitis patients, including the use of regression and machine learning models. Studies involving prediction modeling for tooth loss in periodontitis patients were screened. A search was performed in MEDLINE via PubMed, Embase, and CENTRAL up to 12 February 2022, with citation chasing. Studies exploring model development or external validation studies for models assessing tooth loss in periodontitis patients for clinical use at any time point, with all prediction horizons in English, were considered. Studies were excluded if models were not developed for use in periodontitis patients, were not developed or validated on any data set, predicted outcomes other than tooth loss, or were prognostic factor studies. The CHARMS checklist was used for data extraction, TRIPOD to assess reporting quality, and PROBAST to assess the risk of bias. In total, 4,661 records were screened, and 45 studies were included. Only 26 studies reported any kind of performance measure. The median C-statistic reported was 0.671 (range, 0.57-0.97). All studies were at a high risk of bias due to inappropriate handling of missing data (96%), inappropriate evaluation of model performance (92%), and lack of accounting for model overfitting in evaluating model performance (68%). Many models predicting tooth loss in periodontitis are available, but studies evaluating these models are at a high risk of bias. Model performance measures are likely to be overly optimistic and might not be replicated in clinical use. While this review is unable to recommend any model for clinical practice, it has collated the existing models and their model performance at external validation and their associated sample sizes, which would be helpful to identify promising models for future external validation studies.

Advancements in Text-to-Image Generation through Generative AI

Text-to-image generation, a fascinating intersection of natural language processing and computer vision, has witnessed remarkable progress in recent years. This research paper provides a comprehensive review of the state-of-the-art techniques, challenges, and applications in the field of text-to-image generation. The paper aims to analyze various approaches, discuss their strengths and limitations, and highlight potential directions for future research. Generative Artificial Intelligence (Generative AI) has revolutionized the fusion of textual information and visual content, giving rise to sophisticated Text-to-Image generators. In this context we will discuss dynamic landscape of Generative AI-driven text-to-image synthesis, exploring the state-of-the-art models, underlying architectures, and the impact of training strategies on the quality of generated images. The paper provides a comprehensive overview of Training strategies, encompassing dataset selection and fine-tuning approaches, are scrutinized for their impact on model performance. Common challenges, such as handling ambiguous textual descriptions and ensuring the avoidance of mode collapse, are addressed, offering insights into potential avenues for improvement. Training strategies, encompassing dataset selection and fine-tuning approaches, are scrutinized for their impact on model performance. Common challenges, such as handling ambiguous textual descriptions and ensuring the avoidance of mode collapse, are addressed, offering insights into potential avenues for improvement. Applications of Generative AI text-to-image generators are explored, ranging from content creation to virtual environment design, highlighting their versatility and real-world utility. Ethical considerations surrounding potential misuse, including the creation of deepfakes, are examined to foster a balanced understanding of the technology's societal implications. The research paper concludes with a forward-looking exploration of future directions in Generative AI text-to-image generation. Emphasizing the transformative potential of this technology, the paper envisions advancements in model architectures, training methodologies, and emerging applications, inviting researchers and practitioners to contribute to the ongoing evolution of this exciting field.

Development and validation of a clinical prediction model for glioma grade using machine learning.

Histopathological evaluation is currently the gold standard for grading gliomas; however, this technique is invasive. This study aimed to develop and validate a diagnostic prediction model for glioma by employing multiple machine learning algorithms to identify risk factors associated with high-grade glioma, facilitating the prediction of glioma grading. Data from 1114 eligible glioma patients were obtained from The Cancer Genome Atlas (TCGA) database, which was divided into a training set (n= 781) and a test set (n= 333). Fifty machine learning algorithms were employed, and the optimal algorithm was selected to construct a prediction model. The performance of the machine learning prediction model was compared to the clinical prediction model in terms of discrimination, calibration, and clinical validity to assess the performance of the prediction model. The area under the curve (AUC) values of the machine learning prediction models (training set: 0.870 vs. 0.740, test set: 0.863 vs. 0.718) were significantly improved from the clinical prediction models. Furthermore, significant improvement in discrimination was observed for the Integrated Discrimination Improvement (IDI) (training set: 0.230, test set: 0.270) and Net Reclassification Index (NRI) (training set: 0.170, test set: 0.170) from the clinical prognostic model. Both models showed a high goodness of fit and an increased net benefit. A strong prediction accuracy model can be developed using machine learning algorithms to screen for high-grade glioma risk predictors, which can serve as a non-invasive prediction tool for preoperative diagnostic grading of glioma.

Clinical calculator based on CT and clinicopathologic characteristics predicts short-term prognosis following resection of microsatellite-stabilized diffuse gastric cancer.

Although microsatellite stability/Epithelial-mesenchymal transition (MSS/EMT) subtypes have been reported in multiple cancer prognosis studies, strong confounding factors between MSS/EMT (usually with Lauren's diffuse phenotype) and diffuse gastric cancer (GC) may obscure the independent prognostic value of diffuse GC. Additionally, recent studies suggest a strong correlation between mural stratification based on CT and diffuse GC. This study aims to investigate potential prognostic factors of MSS diffuse GC using mural stratification and to develop a risk assessment model. This retrospective study included 131 patients with MSS diffuse GC who underwent radical surgery. Univariate and multivariate Cox proportional hazards regression analysis was used to identify model predictors and construct a nomogram for overall survival (OS) and recurrence-free survival (RFS) risks. The model's performance was evaluated using ROC, accuracy, and C-index. Internal validation of the model was conducted using the bootstrap resampling method. Among 131 cases, 60 cases (45.8%) exhibited grade 2 mural stratification, which correlated with a poorer tumor prognosis and a more invasive phenotype. Furthermore, a nomogram for predicting OS and RFS prognosis was established based on multivariate results (age, extranodal invasion, mural stratification, and/or P53). The nomogram demonstrated excellent performance, with an AUC of 0.859 (95% CI 0.794-0.924) for OS and 0.859 (95% CI 0.789-0.929) for RFS. Internal validation using 1000 bootstrap samples yielded AUC values of 0.845 and 0.846 for OS and RFS, respectively. Grade 2 mural stratification based on CT imaging revealed a more aggressive invasive phenotype, characterized by increased LN metastasis, higher rates of peritoneal metastasis, and a poorer short-term prognosis. Furthermore, the CT phenotype-based nomogram demonstrates favorable discrimination and calibration, enabling convenient individual short-term prognostic evaluation following resection of MSS diffuse GC.

Longitudinal ultrasound-based AI model predicts axillary lymph node response to neoadjuvant chemotherapy in breast cancer: a multicenter study.

Developing a deep learning radiomics model from longitudinal breast ultrasound and sonographer's axillary ultrasound diagnosis for predicting axillary lymph node (ALN) response to neoadjuvant chemotherapy (NAC) in breast cancer. Breast cancer patients undergoing NAC followed by surgery were recruited from three centers between November 2016 and December 2022. We collected ultrasound images for extracting tumor-derived radiomics and deep learning features, selecting quantitative features through various methods. Two machine learning models based on random forest were developed using pre-NAC and post-NAC features. A support vector machine integrated these data into a fusion model, evaluated via the area under the curve (AUC), decision curve analysis, and calibration curves. We compared the fusion model's performance against sonographer's diagnosis from pre-NAC and post-NAC axillary ultrasonography, referencing histological outcomes from sentinel lymph node biopsy or axillary lymph node dissection. In the validation cohort, the fusion model outperformed both pre-NAC (AUC: 0.899 vs. 0.786, p < 0.001) and post-NAC models (AUC: 0.899 vs. 0.853, p = 0.014), as well as the sonographer's diagnosis of ALN status on pre-NAC and post-NAC axillary ultrasonography (AUC: 0.899 vs. 0.719, p < 0.001). Decision curve analysis revealed patient benefits from the fusion model across threshold probabilities from 0.02 to 0.98. The model also enhanced sonographer's diagnostic ability, increasing accuracy from 71.9% to 79.2%. The deep learning radiomics model accurately predicted the ALN response to NAC in breast cancer. Furthermore, the model will assist sonographers to improve their diagnostic ability on ALN status before surgery. Our AI model based on pre- and post-neoadjuvant chemotherapy ultrasound can accurately predict axillary lymph node metastasis and assist sonographer's axillary diagnosis. Axillary lymph node metastasis status affects the choice of surgical treatment, and currently relies on subjective ultrasound. Our AI model outperformed sonographer's visual diagnosis on axillary ultrasound. Our deep learning radiomics model can improve sonographers' diagnosis and might assist in surgical decision-making.

Development and Validation of an Automated, Real-Time Predictive Model for Postpartum Hemorrhage

OBJECTIVE: To develop and validate a predictive model for postpartum hemorrhage that can be deployed in clinical care using automated, real-time electronic health record (EHR) data and to compare performance of the model with a nationally published risk prediction tool. METHODS: A multivariable logistic regression model was developed from retrospective EHR data from 21,108 patients delivering at a quaternary medical center between January 1, 2018, and April 30, 2022. Deliveries were divided into derivation and validation sets based on an 80/20 split by date of delivery. Postpartum hemorrhage was defined as blood loss of 1,000 mL or more in addition to postpartum transfusion of 1 or more units of packed red blood cells. Model performance was evaluated by the area under the receiver operating characteristic curve (AUC) and was compared with a postpartum hemorrhage risk assessment tool published by the CMQCC (California Maternal Quality Care Collaborative). The model was then programmed into the EHR and again validated with prospectively collected data from 928 patients between November 7, 2023, and January 31, 2024. RESULTS: Postpartum hemorrhage occurred in 235 of 16,862 patients (1.4%) in the derivation cohort. The predictive model included 21 risk factors and demonstrated an AUC of 0.81 (95% CI, 0.79–0.84) and calibration slope of 1.0 (Brier score 0.013). During external temporal validation, the model maintained discrimination (AUC 0.80, 95% CI, 0.72–0.84) and calibration (calibration slope 0.95, Brier score 0.014). This was superior to the CMQCC tool (AUC 0.69 [95% CI, 0.67–0.70], P&lt;.001). The model maintained performance in prospective, automated data collected with the predictive model in real time (AUC 0.82 [95% CI, 0.73–0.91]). CONCLUSION: We created and temporally validated a postpartum hemorrhage prediction model, demonstrated its superior performance over a commonly used risk prediction tool, successfully coded the model into the EHR, and prospectively validated the model using risk factor data collected in real time. Future work should evaluate external generalizability and effects on patient outcomes; to facilitate this work, we have included the model coefficients and examples of EHR integration in the article.

Predicting Mortality Prior to Interhospital Hospital for "Unseen" General Surgery Patients: Development, Validation and Feasibility Trial of a Mortality Risk Calculator.

Develop and validate a mortality risk calculator that could be utilized at the time of transfer, leveraging routinely collected variables that could be obtained by trained non-clinical transfer personnel. There are no objective tools to predict mortality at the time of inter-hospital transfer for Emergency General Surgery (EGS) patients that are "unseen" by the accepting system. Patients transferred to general or colorectal surgery services from January 2016 through August 2022 were retrospectively identified and randomly divided into training and validation cohorts (3:1 ratio). The primary outcome was admission-related mortality, defined as death during the index admission or within 30 days post-discharge. Multiple predictive models were developed and validated. Among 4,664 transferred patients, 280 (6.0%) experienced mortality. Predictive models were generated utilizing 19 routinely collected variables; the penalized regression model was selected over other models due to excellent performance using only 12 variables. The model performance on the validating set resulted in an area under the receiver operating characteristic curve, sensitivity, specificity, and balanced accuracy of 0.851, 0.90, 0.67, and 0.79, respectively. After bias correction, Brier score was 0.04, indicating a strong association between the assigned risk and the observed frequency of mortality. A risk calculator using twelve variables has excellent predictive ability for mortality at the time of interhospital transfer among "unseen" EGS patients. Quantifying a patient's mortality risk at the time of transfer could improve patient triage, bed and resource allocation, and standardize care.

Streamlining Ocean Dynamics Modeling with Fourier Neural Operators: A Multiobjective Hyperparameter and Architecture Optimization Approach

Training an effective deep learning model to learn ocean processes involves careful choices of various hyperparameters. We leverage DeepHyper’s advanced search algorithms for multiobjective optimization, streamlining the development of neural networks tailored for ocean modeling. The focus is on optimizing Fourier neural operators (FNOs), a data-driven model capable of simulating complex ocean behaviors. Selecting the correct model and tuning the hyperparameters are challenging tasks, requiring much effort to ensure model accuracy. DeepHyper allows efficient exploration of hyperparameters associated with data preprocessing, FNO architecture-related hyperparameters, and various model training strategies. We aim to obtain an optimal set of hyperparameters leading to the most performant model. Moreover, on top of the commonly used mean squared error for model training, we propose adopting the negative anomaly correlation coefficient as the additional loss term to improve model performance and investigate the potential trade-off between the two terms. The numerical experiments show that the optimal set of hyperparameters enhanced model performance in single timestepping forecasting and greatly exceeded the baseline configuration in the autoregressive rollout for long-horizon forecasting up to 30 days. Utilizing DeepHyper, we demonstrate an approach to enhance the use of FNO in ocean dynamics forecasting, offering a scalable solution with improved precision.

Eigenvector spatial filtering enhancing natural hazards vulnerability assessment in a susceptible urban environment: A case study of Izmir earthquake in Turkey

The increasing risk of earthquakes in urban areas has made it crucial to develop accurate vulnerability models for city infrastructure and systems. We aimed to assess and compare the effectiveness of different models and vulnerability analysis techniques in predicting earthquake vulnerability in the specific context of Izmir, Turkey. One central hypothesis in this research aimed to determine whether integrating Eigenvector Spatial Filtering (ESF) into both regression models and machine learning algorithms would yield a comparable enhancement in model performance. We performed earthquake vulnerability modeling (EVM) by considering (ⅰ) only seismic-related variables (SRV) and (ⅱ) integrating ESF by using Moran's eigenvector maps (MEMs). For each approach, we evaluated the predictive performance of two simple regression-based models; generalized linear model (GLM) and generalized additive model (GAM), and two complex machine learning ones; generalized boosting model (GBM), and random forest (RF). The study utilized five primary indicators encompassing geotechnical, physical, structural, social, and facilities data. The predictive performance of the models was assessed using evaluation metrics including Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and adjusted R2. The results indicated that the optimal candidate model consisted of five key variables: altitude, building height, distance to safety gathering places, Peak Ground Acceleration (PGA), and population density. We found that decision-tree-based methods performed better than regression-based methods for both modeling schemes. RF exhibited the highest predictive performance for the training data (RMSE = 0.59, adjusted R2 = 0.71), while GBM outperformed other models for the test data (RMSE = 0.79, adjusted R2 = 0.78). However, incorporating ESF to the EVM analysis revealed that regression-based methods, particularly the GLM, obtained highest improvement in accuracy (RMSE 0.94 vs 0.76 and adjusted R2 0.56 vs 0.71 for the SRV and SRV + MEMs modeling approach). Significant differences were observed between GLM-GBM and GLM-RF comparisons, as well as GAM-GBM and GAM-RF comparisons. The findings of this research are expected to be helpful for informed decision-making, targeted risk reduction, and the development of effective policies and strategies to enhance preparedness and resilience in the face of seismic events in highly susceptible urban systems.

Automatic classification and segmentation of blast cells using deep transfer learning and active contours.

Acute lymphoblastic leukemia (ALL) presents a formidable challenge in hematological malignancies, necessitating swift and precise diagnostic techniques for effective intervention. The conventional manual microscopy of blood smears, although widely practiced, suffers from significant limitations including labor-intensity and susceptibility to human error, particularly in distinguishing the subtle differences between normal and leukemic cells. To overcome these limitations, our research introduces the ALLDet classifier, an innovative tool employing deep transfer learning for the automated analysis and categorization of ALL from White Blood Cell (WBC) nuclei images. Our investigation encompassed the evaluation of nine state-of-the-art pre-trained convolutional neural network (CNN) models, namely VGG16, VGG19, ResNet50, ResNet101, DenseNet121, DenseNet201, Xception, MobileNet, and EfficientNetB3. We augmented this approach by incorporating a sophisticated contour-based segmentation technique, derived from the Chan-Vese model, aimed at the meticulous segmentation of blast cell nuclei in blood smear images, thereby enhancing the accuracy of our analysis. The empirical assessment of these methodologies underscored the superior performance of the EfficientNetB3 model, which demonstrated exceptional metrics: a recall specificity of 98.5%, precision of 95.86%, F1-score of 97.16%, and an overall accuracy rate of 97.13%. The Chan-Vese model's adaptability to the irregular shapes of blast cells and its noise-resistant segmentation capability were key to capturing the complex morphological changes essential for accurate segmentation. The combined application of the ALLDet classifier, powered by EfficientNetB3, with our advanced segmentation approach, emerges as a formidable advancement in the early detection and accurate diagnosis of ALL. This breakthrough not only signifies a pivotal leap in leukemia diagnostic methodologies but also holds the promise of significantly elevating the standards of patient care through the provision of timely and precise diagnoses. The implications of this study extend beyond immediate clinical utility, paving the way for future research to further refine and enhance the capabilities of artificial intelligence in medical diagnostics.

Data-Driven Strategies for Complex System Forecasts: The Role of Textual Big Data and State-Space Transformers in Decision Support

In this research, an innovative state space-based Transformer model is proposed to address the challenges of complex system prediction tasks. By integrating state space theory, the model aims to enhance the capability to capture dynamic changes in complex data, thereby improving the accuracy and robustness of prediction tasks. Extensive experimental validations were conducted on three representative tasks, including legal case judgment, legal case translation, and financial data analysis to assess the performance and application potential of the model. The experimental results demonstrate significant performance improvements of the proposed model over traditional Transformer models and other advanced variants such as Bidirectional Encoder Representation from Transformers (BERT) and Finsformer across all evaluated tasks. Specifically, in the task of legal case judgment, the proposed model exhibited a precision of 0.93, a recall of 0.90, and an accuracy of 0.91, significantly surpassing the traditional Transformer model (with precision of 0.78, recall of 0.73, accuracy of 0.76) and performances of other comparative models. In the task of legal case translation, the precision of the proposed model reached 0.95, with a recall of 0.91 and an accuracy of 0.93, also outperforming other models. Likewise, in the task of financial data analysis, the proposed model also demonstrated excellent performance, with a precision of 0.94, recall of 0.90, and accuracy of 0.92. The state space-based Transformer model proposed not only theoretically expands the research boundaries of deep learning models in complex system prediction but also validates its efficiency and broad application prospects through experiments. These achievements provide new insights and directions for future research and development of deep learning models, especially in tasks requiring the understanding and prediction of complex system dynamics.