Deep Learning-based Prediction Research Articles

Machine and deep learning-based prediction of flexural moment capacity of ultra-high performance concrete beams with/out steel fiber

Machine and deep learning-based prediction of flexural moment capacity of ultra-high performance concrete beams with/out steel fiber

A multiview deep learning-based prediction pipeline augmented with confident learning can improve performance in determining knee arthroplasty candidates.

Preoperative prudent patient selection plays a crucial role in knee osteoarthritis management but faces challenges in appropriate referrals such as total knee arthroplasty (TKA), unicompartmental knee arthroplasty (UKA)and nonoperative intervention. Deep learning (DL) techniques can build prediction models for treatment decision-making. The aim is to develop and evaluate a knee arthroplasty prediction pipeline using three-view X-rays to determine the suitable candidates for TKA, UKAor are not arthroplasty candidates. A study was conducted using three-view (anterior-posterior, lateraland patellar) X-rays and surgical data of patients undergoing TKA, UKAor nonarthroplasty interventions from sites A and B. Data from site A were used to derive and validate models. Data from site B were used as external test set. A DL pipeline combining YOLOv3 and ResNet-18 with confident learning (CL) was developed. Multiview Convolutional Neural Network, EfficientNet-b4, ResNet-101and the proposed model without CLwere also trained and tested. The models were evaluated using metrics such as area under the receiver operating characteristic curve (AUC), accuracy, precision, specificity, sensitivityand F1 score. The data set comprised a total of 1779 knees. Of which 1645 knees were from site A as a derivation set and an internal validation cohort. The external validation cohort consisted of 134 knees. The internal validation cohort demonstrated superior performance for the proposed model augmented with CL, achieving an AUC of 0.94 and an accuracy of 85.9%. External validation further confirmed the model's generalisation, with an AUC of 0.93 and an accuracy of 82.1%. Comparative analysis with other neural network models showed the proposed model's superiority. The proposed DL pipeline, integrating YOLOv3, ResNet-18and CL, provides accurate predictions for knee arthroplasty candidates based on three-view X-rays. This prediction model could be useful in performing decision making for the type of arthroplasty procedure in an automated fashion. Level III, diagnostic study.

Analysing and Evaluating the Performance of Deep-Learning-Based Arrhythmia Detection Using Electrocardiogram Signals

Cardiac arrhythmia is a cardiac irregularity that impacts a significant number of individuals globally. Certain arrhythmias may be benign or occur only once, while recurring arrhythmias have the potential to cause organ failure, increase the risk of stroke by a factor of five, and even lead to sudden cardiac death. Thus, in order to identify and treat arrhythmia and prevent potentially fatal cardiac problems, the rapid detection and categorization of Electrocardiogram (ECG) signals is of paramount importance. The non-invasive technique employs electrodes to examine the electrical potentials of the heart, facilitating the detection of structural and functional irregularities that contribute to the diagnosis of cardiovascular illnesses. Nevertheless, the high likelihood of manual interpretation, which is both time-consuming and susceptible to error caused by weariness, poses a significant challenge for cardiologists in identifying and diagnosing cardiac issues. In recent times, there has been a growing utilization of Deep Learning (DL) models in the field of arrhythmia prediction, with the aim of enhancing clinical decision-making and potentially mitigating the likelihood of valetudinarian fatality. These models enhance the diagnostic capabilities of electrocardiography by detecting pathological situations, extracting anatomically meaningful data, evaluating cardio-motion, and assessing the quality of echo images. As a result, they serve as an alternative tool for precise diagnosis and treatment of arrhythmias. In order to forecast and categorize arrhythmias based on ECG signal data, this study provides a comprehensive review of various deep learning approaches. The initial section of the study provides a concise overview of various deep learning-based prediction models that have been developed by multiple researchers for echocardiography systems. Subsequently, a comparative analysis is undertaken to comprehend the limitations of those algorithms and propose a novel approach to improve the accuracy of cardiac view classification in echocardiography systems.

A comprehensive survey on deep learning-based identification and predicting the interaction mechanism of long non-coding RNAs.

Long noncoding RNAs (lncRNAs) have been discovered to be extensively involved in eukaryotic epigenetic, transcriptional, and post-transcriptional regulatory processes with the advancements in sequencing technology and genomics research. Therefore, they play crucial roles in the body's normal physiology and various disease outcomes. Presently, numerous unknown lncRNA sequencing data require exploration. Establishing deep learning-based prediction models for lncRNAs provides valuable insights for researchers, substantially reducing time and costs associated with trial and error and facilitating the disease-relevant lncRNA identification for prognosis analysis and targeted drug development as the era of artificial intelligence progresses. However, most lncRNA-related researchers lack awareness of the latest advancements in deep learning models and model selection and application in functional research on lncRNAs. Thus, we elucidate the concept of deep learning models, explore several prevalent deep learning algorithms and their data preferences, conduct a comprehensive review of recent literature studies with exemplary predictive performance over the past 5years in conjunction with diverse prediction functions, critically analyze and discuss the merits and limitations of current deep learning models and solutions, while also proposing prospects based on cutting-edge advancements in lncRNA research.

Benchmarking machine learning-based real-time respiratory signal predictors in 4D SBRT.

Stereotactic body radiotherapy of thoracic and abdominal tumors has to account for respiratory intrafractional tumor motion. Commonly, an external breathing signal is continuously acquired that serves as a surrogate of the tumor motion and forms the basis of strategies like breathing-guided imaging and gated dose delivery. However, due to inherent system latencies, there exists a temporal lag between the acquired respiratory signal and the system response. Respiratory signal prediction models aim to compensate for the time delays and to improve imaging and dosedelivery. The present study explores and compares six state-of-the-art machine and deep learning-based prediction models, focusing on real-time and real-world applicability. All models and data are provided as open source and data to ensure reproducibility of the results and fosterreuse. The study was based on 2502 breathing signals ( h) acquired during clinical routine, split into independent training (50%), validation (20%), and test sets (30%). Input signal values were sampled from noisy signals, and the target signal values were selected from corresponding denoised signals. A standard linear prediction model (Linear), two state-of-the-art models in general univariate signal prediction (Dlinear, Xgboost), and three deep learning models (Lstm, Trans-Enc, Trans-TSF) were chosen. The prediction performance was evaluated for three different prediction horizons (480, 680, and 920ms). Moreover, the robustness of the different models when applied to atypical, that is, out-of-distribution (OOD) signals, was analyzed. The Lstm model achieved the lowest normalized root mean square error for all prediction horizons. The prediction errors only slightly increased for longer horizons. However, a substantial spread of the error values across the test signals was observed. Compared to typical, that is, in-distribution test signals, the prediction accuracy of all models decreased when applied to OOD signals. The more complex deep learning models Lstm and Trans-Enc showed the least performance loss, while the performance of simpler models like Linear dropped the most. Except for Trans-Enc, inference times for the different models allowed for real-time application. The application of the Lstm model achieved the lowest prediction errors. Simpler prediction filters suffer from limited signal history access, resulting in a drop in performance for OODsignals.

D3EGFR: a webserver for deep learning-guided drug sensitivity prediction and drug response information retrieval for EGFR mutation-driven lung cancer.

As key oncogenic drivers in non-small-cell lung cancer (NSCLC), various mutations in the epidermal growth factor receptor (EGFR) with variable drug sensitivities have been a major obstacle for precision medicine. To achieve clinical-level drug recommendations, a platform for clinical patient case retrieval and reliable drug sensitivity prediction is highly expected. Therefore, we built a database, D3EGFRdb, with the clinicopathologic characteristics and drug responses of 1339 patients with EGFR mutations via literature mining. On the basis of D3EGFRdb, we developed a deep learning-based prediction model, D3EGFRAI, for drug sensitivity prediction of new EGFR mutation-driven NSCLC. Model validations of D3EGFRAI showed a prediction accuracy of 0.81 and 0.85 for patients from D3EGFRdb and our hospitals, respectively. Furthermore, mutation scanning of the crucial residues inside drug-binding pockets, which may occur in the future, was performed to explore their drug sensitivity changes. D3EGFR is the first platform to achieve clinical-level drug response prediction of all approved small molecule drugs for EGFR mutation-driven lung cancer and is freely accessible at https://www.d3pharma.com/D3EGFR/index.php.

Deep Learning Models Compared to Experimental Variability for the Prediction of CYP3A4 Time-Dependent Inhibition.

Most drugs are mainly metabolized by cytochrome P450 (CYP450), which can lead to drug-drug interactions (DDI). Specifically, time-dependent inhibition (TDI) of CYP3A4 isoenzyme has been associated with clinically relevant DDI. To overcome potential DDI issues, high-throughput in vitro assays were established to assess the TDI of CYP3A4 during the discovery and lead optimization phases. However, in silico machine learning models would enable an earlier and larger-scale assessment of TDI potential liabilities. For CYP inhibition, most modeling efforts have focused on highly imbalanced and small data sets. Moreover, assay variability is rarely considered, which is key to understand the model's quality and suitability for decision-making. In this work, machine learning models were built for the prediction of TDI of CYP3A4, evaluated prospectively, and compared to the variability of the experimental assay. Different modeling strategies were investigated to assess their influence on the model's performance. Through multitask learning, additional data sets were leveraged for model building, coming from public databases, in-house CYP-related assays, or other pharmaceutical companies (federated learning). Apart from the numerical prediction of inactivation rates of CYP3A4 TDI, three-class predictions were carried out, giving a negative (inactivation rate kobs < 0.01 min-1), weak positive (0.01 ≤ kobs ≤ 0.025 min-1), or positive (kobs > 0.025 min-1) output. The final multitask graph neural network model achieved misclassification rates of 8 and 7% for positive and negative TDI, respectively. Importantly, the presented deep learning-based predictions had a similar precision to the reproducibility of in vitro experiments and thus offered great opportunities for drug design, early derisk of DDI potential, and selection of experiments. To facilitate CYP inhibition modeling efforts in the public domain, the developed model was used to annotate ∼16 000 publicly available structures, and a surrogate data set is shared as Supporting Information.

PS2MS: A Deep Learning-Based Prediction System for Identifying New Psychoactive Substances Using Mass Spectrometry.

The rapid proliferation of new psychoactive substances (NPS) poses significant challenges to conventional mass-spectrometry-based identification methods due to the absence of reference spectra for these emerging substances. This paper introduces PS2MS, an AI-powered predictive system designed specifically to address the limitations of identifying the emergence of unidentified novel illicit drugs. PS2MS builds a synthetic NPS database by enumerating feasible derivatives of known substances and uses deep learning to generate mass spectra and chemical fingerprints. When the mass spectrum of an analyte does not match any known reference, PS2MS simultaneously examines the chemical fingerprint and mass spectrum against the putative NPS database using integrated metrics to deduce possible identities. Experimental results affirm the effectiveness of PS2MS in identifying cathinone derivatives within real evidence specimens, signifying its potential for practical use in identifying emerging drugs of abuse for researchers and forensic experts.

A systematic pan-cancer study on deep learning-based prediction of multi-omic biomarkers from routine pathology images.

The objective of this comprehensive pan-cancer study is to evaluate the potential of deep learning (DL) for molecular profiling of multi-omic biomarkers directly from hematoxylin and eosin (H&E)-stained whole slide images. A total of 12,093 DL models predicting 4031 multi-omic biomarkers across 32 cancer types were trained and validated. The study included a broad range of genetic, transcriptomic, and proteomic biomarkers, as well as established prognostic markers, molecular subtypes, and clinical outcomes. Here we show that 50% of the models achieve an area under the curve (AUC) of 0.644 or higher. The observed AUC for 25% of the models is at least 0.719 and exceeds 0.834 for the top 5%. Molecular profiling with image-based histomorphological features is generally considered feasible for most of the investigated biomarkers and across different cancer types. The performance appears to be independent of tumor purity, sample size, and class ratio (prevalence), suggesting a degree of inherent predictability in histomorphology. The results demonstrate that DL holds promise to predict a wide range of biomarkers across the omics spectrum using only H&E-stained histological slides of solid tumors. This paves the way for accelerating diagnosis and developing more precise treatments for cancer patients.

Smart meter-based energy consumption forecasting for smart cities using adaptive federated learning

Forecasting short-term residential energy consumption is critical in modern decentralized power systems. Deep learning-based prediction methods that can handle the high variability of residential electrical loads have made models more accurate. On the other hand, these methods need a lot of sensitive information about how much people use something gathered centrally to train a forecasting model. This is not good for privacy and scalability. Moreover, models may become less accurate over time due to changing conditions. In this work, we propose a framework for energy consumption forecasting that exploits adaptive learning, federated learning, and edge computing concepts. A central server aggregates numerous long short-term memory (LSTM) models that users at various locations train with their energy consumption data to create a generalized model that uses adaptive learning to detect data drifts and enhance forecasting at the edge layer. Our findings show that adaptive federated learning performs better than centralized learning while preserving privacy, reducing communication overhead, lowering the forecast error rate by 8%, and decreasing the training time by approximately 80%.

Deep Learning-Based Prediction Model for Gait Recovery after a Spinal Cord Injury.

Predicting gait recovery after a spinal cord injury (SCI) during an acute rehabilitation phase is important for planning rehabilitation strategies. However, few studies have been conducted on this topic to date. In this study, we developed a deep learning-based prediction model for gait recovery after SCI upon discharge from an acute rehabilitation facility. Data were collected from 405 patients with acute SCI admitted to the acute rehabilitation facility of Korea University Anam Hospital between June 2008 and December 2022. The dependent variable was Functional Ambulation Category at the time of discharge (FAC-DC). Seventy-one independent variables were selected from the existing literature: basic information, International Standards for Neurological Classification of SCI scores, neurogenic bladders, initial FAC, and somatosensory-evoked potentials of the lower extremity. Recurrent neural network (RNN), linear regression (LR), Ridge, and Lasso methods were compared for FAC-DC prediction in terms of the root-mean-squared error (RMSE). RNN variable importance, which is the RMSE gap between a complete RNN model and an RNN model excluding a certain variable, was used to evaluate the contribution of this variable. Based on the results of this study, the performance of the RNN was far better than that of LR, Ridge, and Lasso. The respective RMSEs were 0.3738, 2.2831, 1.3161, and 1.0246 for all the participants; 0.3727, 1.7176, 1.3914, and 1.3524 for those with trauma; and 0.3728, 1.7516, 1.1012, and 0.8889 for those without trauma. In terms of RNN variable importance, lower-extremity motor strength (right and left ankle dorsiflexors, right knee extensors, and left long toe extensors) and the neurological level of injury were ranked among the top five across the boards. Therefore, initial FAC was the seventh, third, and ninth most important predictor for all participants, those with trauma, and those without trauma, respectively. In conclusion, this study developed a deep learning-based prediction model with excellent performance for gait recovery after SCI at the time of discharge from an acute rehabilitation facility. This study also demonstrated the strength of deep learning as an explainable artificial intelligence method for identifying the most important predictors.

A novel approach for improving the performance of deep learning-based state of charge estimation of lithium-ion batteries: Choosy SoC Estimator (ChoSoCE)

Deep learning-based (DL) methods have recently come to the forefront among the data-driven models due to their success in capturing the complexities of the battery. Many previous DL-based studies for SoC estimation have almost exclusively focused on improving DL structure by proposing various architectures. Questions regarding the outlier or atypical predictions have yet to be adequately addressed. Furthermore, few works benefit from optimization algorithms to determine the hyperparameters of DL. In this study, we have addressed the problem of how to obtain the hyperparameter of DL and fix the improper DL predictions. To this aim, we used two different optimization algorithms to determine the hyperparameters of DL and proposed a novel algorithm that considers the previous SoC estimations. The algorithm either approves or rejects the DL predictions for the relevant step and offers new values for the rejected ones. The proposed scheme is evaluated using a battery dataset which includes different driving cycles. According to the results, it is observed that the optimized DL outperforms the empirical one by at least 35% in terms of performance indices. Moreover, the proposed novel algorithm successfully integrates into all variations and significantly improves the performance index scores.

A fusion gas load prediction model with three-way residual error amendment

Accurately predicting gas load is crucial for optimal planning and scheduling of natural gas production. Existing machine learning or deep learning-based prediction methods primarily aim to improve accuracy. However, these methods face challenges such as overlapped features, significant background noise, and residual errors. In response to these challenges, we proposed a fusion prediction model, known as the Double Temporal Convolutional Network (TCN) and Light Gradient Boosting Machine (LightGBM) with Three-way Residual Error Amendment (DT-LGBM-3WREA). To address the overlapped feature issue in time series data, we utilize the STL module for data decomposition. To handle background noise, the network incorporates a double TCN module specifically designed for trend and seasonal components. Thus, it can better avoid the transmission of noise between modules. To address residual error, we introduce three-way decision methodology and develop the 3WREA module, which aims to avoid local optima and random factors from perturbing the prediction accuracy. This approach mitigates the influence of remainder item on overall prediction performance. Experimental results demonstrate that the DT-LGBM-3WREA model excels in gas load prediction scenarios, surpassing state-of-the-art methods.

Temporal and spatial satellite data augmentation for deep learning-based rainfall nowcasting

Abstract The significance of improving rainfall prediction methods has escalated due to climate change-induced flash floods and severe flooding. In this study, rainfall nowcasting has been studied utilizing NASA Giovanni satellite-derived precipitation products and the convolutional long short-term memory (ConvLSTM) approach. The goal of the study is to assess the impact of data augmentation on flood nowcasting. Due to data requirements of deep learning-based prediction methods, data augmentation is performed using eight different interpolation techniques. Spatial, temporal, and spatio-temporal interpolated rainfall data are used to conduct a comparative analysis of the results obtained through nowcasting rainfall. This research examines two catastrophic floods that transpired in Türkiye Marmara Region in 2009 and the Central Black Sea Region in 2021, which are selected as the focal case studies. The Marmara and Black Sea regions are prone to frequent flooding, which, due to the dense population, has devastating consequences. Furthermore, these regions exhibit distinct topographical characteristics and precipitation patterns, and the frontal systems that impact them are also dissimilar. The nowcast results for the two regions exhibit a significant difference. Although data augmentation significantly reduced the error values by 59% for one region, it did not yield the same effectiveness for the other region.

Deep learning-based prediction of major arrhythmic events in dilated cardiomyopathy: A proof of concept study.

Prediction of major arrhythmic events (MAEs) in dilated cardiomyopathy represents an unmet clinical goal. Computational models and artificial intelligence (AI) are new technological tools that could offer a significant improvement in our ability to predict MAEs. In this proof-of-concept study, we propose a deep learning (DL)-based model, which we termed Deep ARrhythmic Prevention in dilated cardiomyopathy (DARP-D), built using multidimensional cardiac magnetic resonance data (cine videos and hypervideos and LGE images and hyperimages) and clinical covariates, aimed at predicting and tracking an individual patient's risk curve of MAEs (including sudden cardiac death, cardiac arrest due to ventricular fibrillation, sustained ventricular tachycardia lasting ≥30 s or causing haemodynamic collapse in <30 s, appropriate implantable cardiac defibrillator intervention) over time. The model was trained and validated in 70% of a sample of 154 patients with dilated cardiomyopathy and tested in the remaining 30%. DARP-D achieved a 95% CI in Harrell's C concordance indices of 0.12-0.68 on the test set. We demonstrate that our DL approach is feasible and represents a novelty in the field of arrhythmic risk prediction in dilated cardiomyopathy, able to analyze cardiac motion, tissue characteristics, and baseline covariates to predict an individual patient's risk curve of major arrhythmic events. However, the low number of patients, MAEs and epoch of training make the model a promising prototype but not ready for clinical usage. Further research is needed to improve, stabilize and validate the performance of the DARP-D to convert it from an AI experiment to a daily used tool.

A Deep Long-Term Joint Temporal-Spectral Network for Spectrum Prediction.

Spectrum prediction is a promising technique to release spectrum resources and plays an essential role in cognitive radio networks and spectrum situation generating. Traditional algorithms normally focus on one-dimensional or predict spectrum values in a slot-by-slot manner and thus cannot fully perceive the spectrum states in complex environments and lack timeliness. In this paper, a deep learning-based prediction method with a simple structure is developed for temporal-spectral and multi-slot spectrum prediction simultaneously. Specifically, we first analyze and construct spectrum data suitable for the model to simultaneously achieve long-term and multi-dimensional spectrum prediction. Then, a hierarchical spectrum prediction system is developed that takes advantage of the advanced Bi-ConvLSTM and the seq2seq framework. The Bi-ConvLSTM captures time-frequency characteristics of spectrum data, and the seq2seq framework is used for long-term spectrum prediction. Furthermore, the attention mechanism is used to address the limitations of the seq2seq framework that compresses all inputs into fixed-length vectors, resulting in information loss. Finally, the experimental results have shown that the proposed model has a significant advantage over the benchmark schemes. Especially, the proposed spectrum prediction model achieves 6.15%, 0.7749, 1.0978, and 0.9628 in MAPE, MAE, RMSE, and R2, respectively, which is better than all the baseline deep learning models.

Short-term trajectory prediction for individual metro passengers based on multi-level periodicity mining from semantic trajectory

Short-term trajectory prediction for individual metro passengers is an essential component of many location-based applications, such as bottleneck identification, anomalous trajectory detection, and emergency evacuation. Accurate and interpretable trajectory prediction in short-term remains challenging since the trajectory of individual metro passengers is influenced by time-varying and periodic mobility patterns. Moreover, it is still difficult for deep learning-based prediction approaches to intuitively explain the individual periodic mobility patterns captured by the black-box model. In this study, a long short-term memory method based on multi-level periodicity mining (LSTM-MP) from semantic trajectory is developed to achieve accurate and interpretable short-term trajectory prediction for individual metro passengers. The newly defined semantic trajectories of metro passengers that combine the semantic information about the geometric and operational characteristics of the metro with the raw trajectory data are constructed to help obtain fine-grained periodic mobility patterns of individual passengers. Multi-level periodicity which includes long-range, weekly-interval, and daily-interval periodicities, is calculated with variable time granularity and its polymerization from the constructed semantic trajectories. Finally, the characteristics of multi-level periodicity are considered by developing a principal trajectory attention mechanism to select the most relevant trajectories for prediction in the LSTM-MP framework. The prediction performance of LSTM-MP is compared with 2-order Markov chain and Naïve-LSTM algorithms using the large real-world anonymous passenger WiFi probe dataset. Experimental results show that LSTM-MP can predict when and where an individual will be in the future time series with an average accuracy of 82.3%. This demonstrates the effectiveness of LSTM-MP in the short-term trajectory prediction for individual metro passengers. Potential applications of the short-term trajectory prediction for individual metro passengers are also identified.

Generalisable deep learning method for mammographic density prediction across imaging techniques and self-reported race.

Breast density is an important risk factor for breast cancer complemented by a higher risk of cancers being missed during screening of dense breasts due to reduced sensitivity of mammography. Automated, deep learning-based prediction of breast density could provide subject-specific risk assessment and flag difficult cases during screening. However, there is a lack of evidence for generalisability across imaging techniques and, importantly, across race. This study used a large, racially diverse dataset with 69,697 mammographic studies comprising 451,642 individual images from 23,057 female participants. A deep learning model was developed for four-class BI-RADS density prediction. A comprehensive performance evaluation assessed the generalisability across two imaging techniques, full-field digital mammography (FFDM) and two-dimensional synthetic (2DS) mammography. A detailed subgroup performance and bias analysis assessed the generalisability across participants' race. Here we show that a model trained on FFDM-only achieves a 4-class BI-RADS classification accuracy of 80.5% (79.7-81.4) on FFDM and 79.4% (78.5-80.2) on unseen 2DS data. When trained on both FFDM and 2DS images, the performance increases to 82.3% (81.4-83.0) and 82.3% (81.3-83.1). Racial subgroup analysis shows unbiased performance across Black, White, and Asian participants, despite a separate analysis confirming that race can be predicted from the images with a high accuracy of 86.7% (86.0-87.4). Deep learning-based breast density prediction generalises across imaging techniques and race. No substantial disparities are found for any subgroup, including races that were never seen during model development, suggesting that density predictions are unbiased.

Deep learning-based prediction of saturation magnetic flux density in Fe-based metallic glasses via transfer learning

Transfer learning (TL) is a widely recognized machine learning model that leverages domain knowledge from one or more predictive tasks to enhance performance in a related task. This study demonstrated the improved prediction performance of saturation magnetic flux density (Bs) of bulk metallic glasses (BMGs) with deep learning through the utilization of TL, which enabled the extension of compositional knowledge to overcome the limitations posed by scarce data. Specifically, TL was demonstrated by training a convolutional autoencoder network on full BMG datasets and transferring the acquired compositional knowledge to scarce iron (Fe)-based metallic glass datasets. Through this transfer, the broad compositional knowledge derived from the full BMG dataset is embedded in data representations, allowing its domain knowledge to be shared with smaller datasets. Here, a convolutional neural network was used as a predictor, with the prediction performance of Bs evaluated by the determination coefficient (R2 score) and root-mean-squared error (RMSE). The predictor, having received knowledge about the broad composition through TL, exhibited enhanced predictive performance of Bs on unseen BMG datasets while maintaining its prediction capability for Fe-based metallic glass alloys with R2 = 0.939 and RMSE = 0.084 T. This performance surpassed the performance of directly transferred knowledge on scarce data. These findings underscore the benefits of TL in predictive tasks with limited datasets and highlight the potential for applying this method to the development of new materials, where well-trained models can be repurposed for unlabeled material datasets with desired properties.

Added value of dynamic contrast-enhanced MR imaging in deep learning-based prediction of local recurrence in grade 4 adult-type diffuse gliomas patients

Local recurrences in patients with grade 4 adult-type diffuse gliomas mostly occur within residual non-enhancing T2 hyperintensity areas after surgical resection. Unfortunately, it is challenging to distinguish non-enhancing tumors from edema in the non-enhancing T2 hyperintensity areas using conventional MRI alone. Quantitative DCE MRI parameters such as Ktrans and Ve convey permeability information of glioblastomas that cannot be provided by conventional MRI. We used the publicly available nnU-Net to train a deep learning model that incorporated both conventional and DCE MRI to detect the subtle difference in vessel leakiness due to neoangiogenesis between the non-recurrence area and the local recurrence area, which contains a higher proportion of high-grade glioma cells. We found that the addition of Ve doubled the sensitivity while nonsignificantly decreasing the specificity for prediction of local recurrence in glioblastomas, which implies that the combined model may result in fewer missed cases of local recurrence. The deep learning model predictive of local recurrence may enable risk-adapted radiotherapy planning in patients with grade 4 adult-type diffuse gliomas.