Test Dataset Research Articles

Predicting work ability impairment in post COVID-19 patients: a machine learning model based on clinical parameters

Abstract The Post COVID-19 condition (PCC) is a complex disease affecting health and everyday functioning. This is well reflected by a patient’s inability to work (ITW). In this study, we aimed to investigate factors associated with ITW (1) and to design a machine learning-based model for predicting ITW (2) twelve months after baseline. We selected patients from the post COVID care study (PCC-study) with data on their ability to work. To identify factors associated with ITW, we compared PCC patients with and without ITW. For constructing a predictive model, we selected nine clinical parameters: hospitalization during the acute SARS-CoV-2 infection, WHO severity of acute infection, presence of somatic comorbidities, presence of psychiatric comorbidities, age, height, weight, Karnofsky index, and symptoms. The model was trained to predict ITW twelve months after baseline using TensorFlow Decision Forests. Its performance was investigated using cross-validation and an independent testing dataset. In total, 259 PCC patients were included in this analysis. We observed that ITW was associated with dyslipidemia, worse patient reported outcomes (FSS, WHOQOL-BREF, PHQ-9), a higher rate of preexisting psychiatric conditions, and a more extensive medical work-up. The predictive model exhibited a mean AUC of 0.83 (95% CI: 0.78; 0.88) in the 10-fold cross-validation. In the testing dataset, the AUC was 0.76 (95% CI: 0.58; 0.93). In conclusion, we identified several factors associated with ITW. The predictive model performed very well. It could guide management decisions and help setting mid- to long-term treatment goals by aiding the identification of patients at risk of extended ITW.

NNFit: A Self-Supervised Deep Learning Method for Accelerated Quantification of High- Resolution Short Echo Time MR Spectroscopy Datasets.

"Just Accepted" papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content. Purpose To develop and evaluate the performance of NNFit, a self-supervised deep-learning method for quantification of high-resolution short echo-time (TE) echo-planar spectroscopic imaging (EPSI) datasets, with the goal of addressing the computational bottleneck of conventional spectral quantification methods in the clinical workflow. Materials and Methods This retrospective study included 89 short-TE whole-brain EPSI/GRAPPA scans from clinical trials for glioblastoma (Trial 1, May 2014-October 2018) and major-depressive-disorder (Trial 2, 2022- 2023). The training dataset included 685k spectra from 20 participants (60 scans) in Trial 1. The testing-dataset included 115k spectra from 5 participants (13 scans) in Trial 1 and 145k spectra from 7 participants (16 scans) in Trial 2. A comparative analysis was performed between NNFit and a widely used parametric-modeling spectral quantitation method (FITT). Metabolite maps generated by each method were compared using the structural- similarity-index-measure (SSIM) and linear-correlation-coefficient (R2). Radiation treatment volumes for glioblastoma based on the metabolite maps were compared with the Dice-coefficient and a two-tailed t test. Results Average SSIM and R2 scores for Trial 1 test set data were 0.91/0.90 (choline), 0.93/0.93 (creatine), 0.93/0.93 (n-acetylaspartate), 0.80/0.72 (myo-inositol), and 0.59/0.47 (glutamate + glutamine). Average scores for Trial 2 test set data were 0.95/0.95, 0.98/0.97, 0.98/0.98, 0.92/0.92, and 0.79/0.81 respectively. The treatment volumes had average Dice coefficient of 0.92. NNFit's average processing time was 90.1 seconds, whereas FITT took 52.9 minutes on average. Conclusion This study demonstrates that a deep learning approach to spectral quantitation offers comparable performance to conventional quantification methods for EPSI data, but with faster processing at short-TE. ©RSNA, 2025.

Artificial intelligence as a transforming factor in motility disorders–automatic detection of motility patterns in high-resolution anorectal manometry

High-resolution anorectal manometry (HR-ARM) is the gold standard for anorectal functional disorders’ evaluation, despite being limited by its accessibility and complex data analysis. The London Protocol and Classification were developed to standardize anorectal motility patterns classification. This proof-of-concept study aims to develop and validate an artificial intelligence model for identification and differentiation of disorders of anal tone and contractility in HR-ARM. A dataset of 701 HR-ARM exams from a tertiary center, classified according to London Classification, was used to develop and test multiple machine learning (ML) algorithms. The exams were divided in a training and testing dataset with a 80/20% ratio. The testing dataset was used for models’ evaluation through its accuracy, sensitivity, specificity, positive and negative predictive values and area under the receiving-operating characteristic curve. LGBM Classifier had the best performance, with an accuracy of 87.0% for identifying disorders of anal tone and contractility. Different ML models excelled in distinguishing specific disorders of anal tone and contractility, with accuracy over 90.0%. This is the first worldwide study proving the accuracy of a ML model for differentiation of motility patterns in HR-ARM, demonstrating the value of artificial intelligence models in optimizing HR-ARM availability while reducing interobserver variability and increasing accuracy.

Unsupervised Semantic Segmentation of Urban Scenes via Cross-Modal Distillation

Abstract Semantic image segmentation models typically require extensive pixel-wise annotations, which are costly to obtain and prone to biases. Our work investigates learning semantic segmentation in urban scenes without any manual annotation. We propose a novel method for learning pixel-wise semantic segmentation using raw, uncurated data from vehicle-mounted cameras and LiDAR sensors, thus eliminating the need for manual labeling. Our contributions are as follows. First, we develop a novel approach for cross-modal unsupervised learning of semantic segmentation by leveraging synchronized LiDAR and image data. A crucial element of our method is the integration of an object proposal module that examines the LiDAR point cloud to generate proposals for spatially consistent objects. Second, we demonstrate that these 3D object proposals can be aligned with corresponding images and effectively grouped into semantically meaningful pseudo-classes. Third, we introduce a cross-modal distillation technique that utilizes image data partially annotated with the learnt pseudo-classes to train a transformer-based model for semantic image segmentation. Fourth, we demonstrate further significant improvements of our approach by extending the proposed model using a teacher-student distillation with an exponential moving average and incorporating soft targets from the teacher. We show the generalization capabilities of our method by testing on four different testing datasets (Cityscapes, Dark Zurich, Nighttime Driving, and ACDC) without any fine-tuning. We present an in-depth experimental analysis of the proposed model including results when using another pre-training dataset, per-class and pixel accuracy results, confusion matrices, PCA visualization, k-NN evaluation, ablations of the number of clusters and LiDAR’s density, supervised finetuning as well as additional qualitative results and their analysis.

Efficient and accurate commissioning and quality assurance of radiosurgery beam via prior-embedded implicit neural representation learning.

Dosimetric commissioning and quality assurance (QA) for linear accelerators (LINACs) present a significant challenge for clinical physicists due to the high measurement workload and stringent precision standards. This challenge is exacerbated for radiosurgery LINACs because of increased measurement uncertainty and more demanding setup accuracy for small-field beams. Optimizing physicists' effort during beam measurements while ensuring the quality of the measured data is crucial for clinical efficiency and patient safety. To develop a radiosurgery LINAC beam model that embeds prior knowledge of beam data through implicit neural representation (NeRP) learning and to evaluate the model's effectiveness in guiding beam data sampling, predicting complete beam dataset from sparse samples, and verifying detector choice and setup during commissioning and QA. Beam data including lateral profile and tissue-phantom-ratio (TPR), collected from CyberKnife LINACs, were investigated. Multi-layer perceptron (MLP) neural networks were optimized to parameterize a continuous function of the beam data, implicitly defined by the mapping from measurement coordinates to measured dose values. Beam priors were embedded into network weights by first training the network to learn the NeRP of a vendor-provided reference dataset. The prior-embedded network was further fine-tuned with sparse clinical measurements and used to predict unacquired beam data. Prospective and retrospective evaluations of different beam data samples in finetuning the model were performed using the reference beam dataset and clinical testing datasets, respectively. Model prediction accuracy was evaluated over 10 clinical datasets collected from various LINACs with different manufacturing modes and collimation systems. Model sensitivity in detecting beam data acquisition errors including inaccurate detector positioning and inappropriate detector choice was evaluated using two additional datasets with intentionally introduced erroneous samples. Prospective and retrospective evaluations identified consistent beam data samples that are most effective in fine-tuning the model for complete beam data prediction. Despite of discrepancies between clinical beam and the reference beam, fine-tuning the model with sparse beam profile measured at a single depth or with beam TPR measured at a single collimator size predicted beam data that closely match ground truth water tank measurements. Across the 10 clinical beam datasets, the averaged mean absolute error (MAE) in percentage dose was lower than 0.5% and the averaged 1D Gamma passing rate (1%/0.5mm for profile and 1%/1mm for TPR) was higher than 99%. In contrast, the MAE and Gamma passing rates were above 1% and below 95% between the reference beam dataset and clinical beam datasets. Model sensitivity to beam data acquisition errors was demonstrated by significant model prediction changes when fine-tuned with erroneous versus correct beam data samples, as quantified by a Gamma passing rate as low as 18.16% between model predictions. A model for small-field radiosurgery beam was proposed that embeds prior knowledge of beam properties and predicts the entire beam data from sparse measurements. The model can serve as a valuable tool for clinical physicists to verify the accuracy of beam data acquisition and promises to improve commissioning and QA reliability and efficiency with substantially reduced number of beam measurements.

Deep Radiogenomics Sequencing for Breast Tumor Gene-Phenotype Decoding Using Dynamic Contrast Magnetic Resonance Imaging

Abstract Purpose We aim to perform radiogenomic profiling of breast cancer tumors using dynamic contrast magnetic resonance imaging (MRI) for the estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) genes. Methods The dataset used in the current study consists of imaging data of 922 biopsy-confirmed invasive breast cancer patients with ER, PR, and HER2 gene mutation status. Breast MR images, including a T1-weighted pre-contrast sequence and three post-contrast sequences, were enrolled for analysis. All images were corrected using N4 bias correction algorithms. Based on all images and tumor masks, a bounding box of 128 × 128 × 68 was chosen to include all tumor regions. All networks were implemented in 3D fashion with input sizes of 128 × 128 × 68, and four images were input to each network for multi-channel analysis. Data were randomly split into train/validation (80%) and test set (20%) with stratification in class (patient-wise), and all metrics were reported in 20% of the untouched test dataset. Results For ER prediction, SEResNet50 achieved an AUC mean of 0.695 (CI95%: 0.610–0.775), a sensitivity of 0.564, and a specificity of 0.787. For PR prediction, ResNet34 achieved an AUC mean of 0.658 (95% CI: 0.573–0.741), a sensitivity of 0.593, and a specificity of 0.734. For HER2 prediction, SEResNext101 achieved an AUC mean of 0.698 (95% CI: 0.560–0.822), a sensitivity of 0.750, and a specificity of 0.625. Conclusion The current study demonstrated the feasibility of imaging gene-phenotype decoding in breast tumors using MR images and deep learning algorithms with moderate performance.

Estimation of barrier layer thickness from satellite-derived sea surface parameters in the north Indian Ocean using machine learning

ABSTRACT Oceanic barrier layer thickness (BLT) has been estimated from sea surface parameters through four machine learning (ML) models. The BLT used for training and testing of ML models has been computed from 19 years (2005–2023) of Argo data of vertical temperature and salinity profile in the north Indian ocean. It is observed that thick barrier layer (BL) exists in three regions within the north Indian ocean: the Bay of Bengal (BoB), Eastern Arabian Sea (EAS) and the East Equatorial Indian Ocean (EEIO) off the Sumatra. The observed BLT ranges from 0 to 80 m in the north Indian ocean with a seasonal cycle. Four ML models, i.e. multiple linear regression, artificial neural network (ANN), Random Forest and XGBoost have been trained for 2010–2018 to learn the relationship between sea surface parameters (input features) and BLT (target/output variable) in three thick BL regions within the north Indian Ocean. Thereafter, the trained models have been tested for the estimation of BLT in the same regions for the time period of 2019–2023. The ML models are well able to estimate the BLT from sea surface parameters (SST, SSS) with RMSE less than 5 m in the BoB and EAS, which is less than 8 percentile of the model testing datasets. The ‘Random Forest Regressor’ turns out to perform better than the other three ML models for estimation of BLT. Further, the Random Forest estimated BLT has been compared with the BLT computed using ORAS5 reanalysis data of vertical temperature and salinity profile. This study explores the capability of the ML models for capturing the signature of a subsurface variable on the surface owing to the limitation of costly subsurface measurement.

An Efficient 3D Convolutional Neural Network for Dose Prediction in Cancer Radiotherapy from CT Images

Introduction: Cancer is a highly lethal disease with a significantly high mortality rate. One of the most commonly used methods for treatment is radiation therapy. However, cancer treatment using radiotherapy is a time-consuming process that requires significant manual work from planners and doctors. In radiation therapy treatment planning, determining the dose distribution for each of the regions of the patient’s body is one of the most difficult and important tasks. Nowadays, artificial intelligence has shown promising results in improving the quality of disease treatment, particularly in cancer radiation therapy. Objectives: The main objective of this study is to build a high-performance deep learning model for predicting radiation therapy doses for cancer and to develop software to easily manipulate and use this model. Materials and Methods: In this paper, we propose a custom 3D convolutional neural network model with a U-Net-based architecture to automatically predict radiation doses during cancer radiation therapy from CT images. To ensure that the predicted doses do not have negative values, which are not valid for radiation doses, a rectified linear unit (ReLU) function is applied to the output to convert negative values to zero. Additionally, a proposed loss function based on a dose–volume histogram is used to train the model, ensuring that the predicted dose concentrations are highly meaningful in terms of radiation therapy. The model is developed using the OpenKBP challenge dataset, which consists of 200, 100, and 40 head and neck cancer patients for training, testing, and validation, respectively. Before the training phase, preprocessing and augmentation techniques, such as standardization, translation, and flipping, are applied to the training set. During the training phase, a cosine annealing scheduler is applied to update the learning rate. Results and Conclusions: Our model achieved strong performance, with a good DVH score (1.444 Gy) on the test dataset, compared to previous studies and state-of-the-art models. In addition, we developed software to display the dose maps predicted by the proposed model for each 2D slice in order to facilitate usage and observation. These results may help doctors in treating cancer with radiation therapy in terms of both time and effectiveness.

Analysis of borehole strain anomalies before the 2017 Jiuzhaigou Ms 7.0 earthquake based on a graph neural network

Abstract. On 8 August 2017, a strong earthquake of magnitude 7.0 occurred in Jiuzhaigou, Sichuan Province, China. To assess pre-earthquake anomalies, we utilized variational mode decomposition to preprocess borehole strain observation data and combined them with a graph WaveNet neural network model to process data from multiple stations. We obtained 1-year data from four stations near the epicenter as the training dataset and data from 1 January to 10 August 2017 as the test dataset. For the prediction results of the variational mode decomposition–graph WaveNet model, the anomalous days were extracted using statistical methods, and the results of anomalous-day accumulation at multiple stations showed that an increase in the number of anomalous days occurred 15–32 d before the earthquake. The acceleration effect of anomalous accumulation was most obvious 20 d before the earthquake, and an increase in the number of anomalous days also occurred in the 1 to 3 d post-earthquake. We tentatively deduce that the pre-earthquake anomalies are caused by the diffusion of strain energy near the epicenter during the accumulation process, which can be used as a signal of pre-seismic anomalies, whereas the post-earthquake anomalies are caused by the frequent occurrence of aftershocks.

New Library-Based Methods for Nontargeted Compound Identification by GC-EI-MS.

While gas chromatography mass spectrometry (GC-MS) has long been used to identify compounds in complex mixtures, this process is often subjective and time-consuming and leaves a large fraction of seemingly good-quality spectra unidentified. In this work, we describe a set of new mass spectral library-based methods to assist compound identification in complex mixtures. These methods employ mass spectral uniqueness and compound ubiquity of library entries alongside noise reduction and automated comparison of retention indices to library compounds. As a test data set, we used a publicly available electron ionization mass spectrometry data set consisting of 4833 spectra of particulate organic compounds emitted by combustion of wildland fuels. In the present work, spectra in this data set were first identified using the NIST 2023 EI-MS Library and associated batch process identification software (NIST MS PepSearch) using retention-index corrected Identity Search scoring. Resulting identifications and related information were then employed to parametrize other factors that correlate with identification. A method for identifying compounds absent from but related to those present in mass spectral libraries using the Hybrid Similarity Search is illustrated. Nevertheless, some 90% of the spectra remain unidentified. Through comparison of unidentified to identified mass spectra in this data set, a new simple measure, namely median relative abundance, was developed for evaluating the likelihood of identification.

Efficient optical trapping force tuning for cusp-catastrophe autofocusing beams using deep neural networks

Structured light adjusts optical trapping forces through flexible structure design. However, it is challenging to evaluate optical forces on microscopic particles in structured light due to high computational hardware requirements, prolonged computation times, and data inefficiencies associated with solving optical trapping forces using generalized Lorenz–Mie theory. We propose the use of deep neural networks for predicting and tuning the optical trapping force of cusp-catastrophe autofocusing beams on Mie particles. Inputs include beam's structural parameters, laser power, and the size of captured particle, while the output is the optical trapping force. Following iterative training, the neural network achieved a mean square error of 1.5×10−5. Evaluation using 150 sets of test data revealed that 95.3% of the predictions had a relative error of less than 1.8%, indicating a high prediction accuracy. In contrast to traditional computational methods, the neural network model demonstrates a remarkable efficiency improvement—104 times faster in optimizing beams for optical trapping. This advancement demonstrates the advantage of deep learning neural networks for the application of structured light including autofocusing beams in optical tweezers.

Supervised machine learning statistical models for visual outcome prediction in macular hole surgery: a single-surgeon, standardized surgery study

PurposeTo evaluate the predictive accuracy of various machine learning (ML) statistical models in forecasting postoperative visual acuity (VA) outcomes following macular hole (MH) surgery using preoperative optical coherence tomography (OCT) parameters.MethodsThis retrospective study included 158 eyes (151 patients) with full-thickness MHs treated between 2017 and 2023 by the same surgeon and using the same intraoperative surgical technique. Data from electronic medical records and OCT scans were extracted, with OCT-derived qualitative and quantitative MH characteristics recorded. Six supervised ML models—ANCOVA, Random Forest (RF) regression, K-Nearest Neighbor, Support Vector Machine, Extreme Gradient Boosting, and Lasso regression—were trained using an 80:20 training-to-testing split. Model performance was evaluated on an independent testing dataset using the XLSTAT software. In total, the ML statistical models were trained and tested on 14,652 OCT data points from 1332 OCT images.ResultsOverall, 91% achieved MH closure post-surgery, with a median VA gain of -0.3 logMAR units. The RF regression model outperformed other ML models, achieving the lowest mean square error (MSE = 0.038) on internal validation. The most significant predictors of VA were postoperative MH closure status (variable importance = 43.078) and MH area index (21.328). The model accurately predicted the post-operative VA within 0.1, 0.2 and 0.3 logMAR units in 61%, 78%, and 87% of OCT images, respectively.ConclusionThe RF regression model demonstrated superior predictive accuracy for forecasting postoperative VA, suggesting ML-driven approaches may improve surgical planning and patient counselling by providing reliable insights into expected visual outcomes based on pre-operative OCT features.Clinical trial registration numberNot applicable.

Gene‐expression programming‐based models for comprehensive evaluation of bond strength and critical slip of various types of FRP bars in concrete

AbstractThis study aims to present a novel approach to assess the bond behavior of various types of FRP bars in concrete, based on an extensive dataset. The novelty of this research lies in evaluating both the bond strength and the critical slip, which forms the basis for creating a bond model between the bars and concrete. Furthermore, the model is constructed using advanced machine learning techniques, specifically gene‐expression programming (GEP), and its reliability is ensured by relying on a more extensive and diverse experimental dataset compared to previous studies. To train and validate the GEP model, data were collected from experiments involving 793 test specimens, categorized into five groups based on surface treatments: smooth (Sm), sand‐coated (SC), helically wrapped (HW), grooved (Gr), and a hybrid surface (Hbr) achieved by combining SC and HW. The model also considers a wide range of other parameters, including compressive strength of concrete and diameter, tensile strength, elastic modulus, and embedment length of FRP bars. The verification results using a separate testing dataset demonstrate that the proposed models accurately predict the bond strength and critical slip of FRP bars in concrete, achieving R2 values of 0.794 and 0.88, respectively. Comprehensive parametric analyses are performed to clarify the influence of key parameters on the bond and slip behavior of FRP bars in concrete. The analysis results confirm the critical role of the surface type and elastic modulus in enhancing bond strength; particularly, among the different surface treatment methods, the HW treatment exhibits the highest bond strength, followed by the HBr, Gr, and SC surface types.

High-risk habitat radiomics model based on ultrasound images for predicting lateral neck lymph node metastasis in differentiated thyroid cancer

BackgroundThis study aims to evaluate the predictive usefulness of a habitat radiomics model based on ultrasound images for anticipating lateral neck lymph node metastasis (LLNM) in differentiated thyroid cancer (DTC), and for pinpointing high-risk habitat regions and significant radiomics traits.MethodsA group of 214 patients diagnosed with differentiated thyroid carcinoma (DTC) between August 2021 and August 2023 were included, consisting of 107 patients with confirmed postoperative lateral lymph node metastasis (LLNM) and 107 patients without metastasis or lateral cervical lymph node involvement. An additional cohort of 43 patients was recruited to serve as an independent external testing group for this study. Patients were randomly divided into training and internal testing group at an 8:2 ratio. Region of interest (ROI) was manually outlined, and habitat analysis subregions were defined using the K-means method. The ideal number of subregions (n = 5) was determined using the Calinski-Harabasz score, leading to the creation of a habitat radiomics model with 5 subregions and the identification of the high-risk habitat model. Area under the curve (AUC) values were calculated for all models to assess their validity, and predictive model nomograms were created by integrating clinical features. The internal and external testing dataset is employed to assess the predictive performance and stability of the model.ResultsIn internal testing group, Habitat 3 was identified as the high-risk habitat model in the study, showing the best diagnostic efficacy among all models (AUC(CRM) vs. AUC(Habitat 3) vs. AUC(CRM + Habitat 3) = 0.84(95%CI:0.71–0.97) vs. 0.90(95%CI:0.80-1.00) vs. 0.79(95%CI:0.65–0.93)). Moreover, integrating the Habitat 3 model with clinical features and constructing nomograms enhanced the predictive capability of the combined model (AUC = 0.95(95%CI:0.88-1.00)). In this study, an independent external testing cohort was utilized to assess the model’s accuracy, yielding an AUC of 0.88 (95%CI: 0.78–0.98).ConclusionThe integration of the High-Risk Habitats (Habitat 3) radiomics model with clinical characteristics demonstrated a high predictive accuracy in identifying LLNM. This model has the potential to offer valuable guidance to surgeons in deciding the necessity of LLNM dissection for DTC.Clinical trial numberNot applicable.

Investigating LiDAR Metrics for Old-Growth Beech- and Spruce-Dominated Forest Identification in Central Europe

Old-growth forests are essential for maintaining biodiversity, as they are formed by the complexity of diverse forest structures, such as broad variations in tree height and diameter (DBH) and conditions of living and dead trees, leading to various ecological niches. However, many efforts of old-growth forest mapping from LiDAR have targeted only one specific forest structure (e.g., stand height, basal area, or stand density) by deriving information through a large number of LiDAR metrics. This study introduces a novel approach for identifying old-growth forests by optimizing a set of selected LiDAR standards and structural metrics. These metrics effectively capture the arrangement of multiple forest structures, such as canopy heterogeneity, multilayer canopy profile, and canopy openness. To determine the important LiDAR standard and structural metrics in identifying old-growth forests, multicollinearity analysis using the variance inflation factor (VIF) approach was applied to identify and remove metrics with high collinearity, followed by the random forest algorithm to rank which LiDAR standard and structural metrics are important in old-growth forest classification. The results demonstrate that the LiDAR structural metrics (i.e., advanced LiDAR metrics related to multiple canopy structures) are more important and effective in distinguishing old- and second-growth forests than LiDAR standard metrics (i.e., height- and density-based LiDAR metrics) using the European definition of a 150-year stand age threshold for old-growth forests. These structural metrics were then used as predictors for the final classification of old-growth forests, yielding an overall accuracy of 78%, with a true skill statistic (TSS) of 0.58 for the test dataset. This study demonstrates that using a few structural LiDAR metrics provides more information than a high number of standard LiDAR metrics, particularly for identifying old-growth forests in mixed temperate forests. The findings can aid forest and national park managers in developing a practical and efficient old-growth forest identification and monitoring method using LiDAR.

Generation of high-resolution MPRAGE-like images from 3D head MRI localizer (AutoAlign Head) images using a deep learning-based model.

Magnetization prepared rapid gradient echo (MPRAGE) is a useful three-dimensional (3D) T1-weighted sequence, but is not a priority in routine brain examinations. We hypothesized that converting 3D MRI localizer (AutoAlign Head) images to MPRAGE-like images with deep learning (DL) would be beneficial for diagnosing and researching dementia and neurodegenerative diseases. We aimed to establish and evaluate a DL-based model for generating MPRAGE-like images from MRI localizers. Brain MRI examinations including MPRAGE taken at a single institution for investigation of mild cognitive impairment, dementia and epilepsy between January 2020 and December 2022 were included retrospectively. Images taken in 2020 or 2021 were assigned to training and validation datasets, and images from 2022 were used for the test dataset. Using the training and validation set, we determined one model using visual evaluation by radiologists with reference to image quality metrics of peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and Learned Perceptual Image Patch Similarity (LPIPS). The test dataset was evaluated by visual assessment and quality metrics. Voxel-based morphometric analysis was also performed, and we evaluated Dice score and volume differences between generated and original images of major structures were calculated as absolute symmetrized percent change. Training, validation, and test datasets comprised 340 patients (mean age, 56.1 ± 24.4years; 195 women), 36 patients (67.3 ± 18.3years, 20 women), and 193 patients (59.5 ± 24.4years; 111 women), respectively. The test dataset showed: PSNR, 35.4 ± 4.91; SSIM, 0.871 ± 0.058; and LPIPS 0.045 ± 0.017. No overfitting was observed. Dice scores for the segmentation of main structures ranged from 0.788 (left amygdala) to 0.926 (left ventricle). Quadratic weighted Cohen kappa values of visual score for medial temporal lobe between original and generated images were 0.80-0.88. Images generated using our DL-based model can be used for post-processing and visual evaluation of medial temporal lobe atrophy.

Airbnb pricing in Sydney: predictive modelling and explainable machine learning

ABSTRACT We employ multiple predictive algorithms combined with explainable machine learning techniques to forecast and interpret Airbnb rental prices in Sydney, Australia. The best-performing model is selected using multiple metrics and model confidence sets from a variety of methods ranging from simple linear regression to more complex forecast combinations. In addition, we evaluate the importance of feature engineering by training the models on datasets constructed with and without feature engineering and assessing their respective accuracies. Ensemble methods, particularly stacking regressions, outperform other algorithms on both the training and test datasets, while linear models perform the worst. Factors such as property capacity, proximity to popular areas and luxury amenities increase price predictions according to Shapley values, whereas being near major highway entrances is linked to lower prices, likely due to noise and air pollution.

Design of a Novel Fractional Whale Optimization-Enhanced Support Vector Regression (FWOA-SVR) Model for Accurate Solar Energy Forecasting

This study presents a novel Fractional Whale Optimization Algorithm-Enhanced Support Vector Regression (FWOA-SVR) framework for solar energy forecasting, addressing the limitations of traditional SVR in modeling complex relationships within data. The proposed framework incorporates fractional calculus in the Whale Optimization Algorithm (WOA) to improve the balance between exploration and exploitation during hyperparameter tuning. The FWOA-SVR model is comprehensively evaluated against traditional SVR, Long Short-Term Memory (LSTM), and Backpropagation Neural Network (BPNN) models using training, validation, and testing datasets. Experimental results show that FWOA-SVR achieves superior performance with the lowest MSE values (0.036311, 0.03942, and 0.03825), RMSE values (0.19213, 0.19856, and 0.19577), and the highest R2 values (0.96392, 0.96104, and 0.96192) for training, validation, and testing, respectively. These results highlight the significant improvements of FWOA-SVR in prediction accuracy and efficiency, surpassing benchmark models in capturing complex patterns within the data. The findings highlight the effectiveness of integrating fractional optimization techniques into machine learning frameworks for advancing solar energy forecasting solutions.

AUTOMATED OPPORTUNISTIC CT-SCREENING OF ABDOMINAL AORTIC ANEURYSMS USING ARTIFICIAL INTELLIGENCE: PROSPECTS AND CHALLENGES (LITERATURE REVIEW)

Highlights Non-contrast computed tomography (CT) scan is a promising modality for opportunistic screening of abdominal aortic aneurysm (AAA).Automation of opportunistic screening of AAA according to CT data is a promising use of artificial intelligence (AI) technologies.The development of AI algorithms for opportunistic screening of AAA based on CT data is currently limited due to the high labor costs in preparing datasets for AI training and testing. AnnotationAbdominal aortic aneurysm is a cardiovascular disease characterized by a latent progression and adverse prognosis. Timely diagnosis reduces surgical risks and postoperative complications. Diagnostic imaging methods used to detect and evaluate this disease, particularly in targeted and opportunistic screening, are reviewed. The prospects of automation using artificial intelligence technologies for opportunistic screening are explored, moreover, they have already proven to be effective tools for optimizing radiology reports in several fields. This review highlights, however, relatively poor development of artificial intelligence algorithms for opportunistic screening of abdominal aortic aneurysms on native non-contrast abdominal CT studies. Possible reasons for this phenomenon and potential ways of development of this subject area are investigated.

UniAMP: enhancing AMP prediction using deep neural networks with inferred information of peptides

Antimicrobial peptides (AMPs) have been widely recognized as a promising solution to combat antimicrobial resistance of microorganisms due to the increasing abuse of antibiotics in medicine and agriculture around the globe. In this study, we propose UniAMP, a systematic prediction framework for discovering AMPs. We observe that feature vectors used in various existing studies constructed from peptide information, such as sequence, composition, and structure, can be augmented and even replaced by information inferred by deep learning models. Specifically, we use a feature vector with 2924 values inferred by two deep learning models, UniRep and ProtT5, to demonstrate that such inferred information of peptides suffice for the task, with the help of our proposed deep neural network model composed of fully connected layers and transformer encoders for predicting the antibacterial activity of peptides. Evaluation results demonstrate superior performance of our proposed model on both balanced benchmark datasets and imbalanced test datasets compared with existing studies. Subsequently, we analyze the relations among peptide sequences, manually extracted features, and automatically inferred information by deep learning models, leading to observations that the inferred information is more comprehensive and non-redundant for the task of predicting AMPs. Moreover, this approach alleviates the impact of the scarcity of positive data and demonstrates great potential in future research and applications.