Bioinformatics. 2024 Aug 28:btae528. doi: 10.1093/bioinformatics/btae528. Online ahead of print.

ABSTRACT

MOTIVATION: 5-hydroxymethylcytosine (5hmC), a crucial epigenetic mark with a significant role in regulating tissue-specific gene expression, is essential for understanding the dynamic functions of the human genome. Despite its importance, predicting 5hmC modification across the genome remains a challenging task, especially when considering the complex interplay between DNA sequences and various epigenetic factors such as histone modifications and chromatin accessibility.

RESULTS: Using tissue-specific 5hmC sequencing data, we introduce Deep5hmC, a multimodal deep learning framework that integrates both the DNA sequence and epigenetic features such as histone modification and chromatin accessibility to predict genome-wide 5hmC modification. The multimodal design of Deep5hmC demonstrates remarkable improvement in predicting both qualitative and quantitative 5hmC modification compared to unimodal versions of Deep5hmC and state-of-the-art machine learning methods. This improvement is demonstrated through benchmarking on a comprehensive set of 5hmC sequencing data collected at four developmental stages during forebrain organoid development and across 17 human tissues. Compared to DeepSEA and random forest, Deep5hmC achieves close 4 % and 17% improvement of AUROC across four forebrain developmental stages, and 6% and 27% across 17 human tissues for predicting binary 5hmC modification sites; and 8% and 22% improvement of Spearman correlation coefficient across four forebrain developmental stages, and 17% and 30% across 17 human tissues for predicting continuous 5hmC modification. Notably, Deep5hmC showcases its practical utility by accurately predicting gene expression and identifying differentially hydroxymethylated regions in a case-control study of Alzheimer's disease. Deep5hmC significantly improves our understanding of tissue-specific gene regulation and facilitates the development of new biomarkers for complex diseases.

AVAILABILITY AND IMPLEMENTATION: Deep5hmC is available via https://github.com/lichen-lab/Deep5hmC.

SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

PMID:39196755 | DOI:10.1093/bioinformatics/btae528

IEEE J Biomed Health Inform. 2024 Aug 28;PP. doi: 10.1109/JBHI.2024.3451331. Online ahead of print.

ABSTRACT

Recent studies have identified microvascular invasion (MVI) as the most vital independent biomarker associated with early tumor recurrence. With advancements in medical technology, several computational methods have been developed to predict preoperative MVI using diverse medical images. These existing methods rely on human experience, attribute selection or clinical trial testing, which is often time-consuming and labor-intensive. Leveraging the advantages of deep learning, this study presents a novel end-to-end algorithm for predicting MVI prior to surgery. We devised a series of data preprocessing strategies to fully extract multi-view features from the data while preserving peritumoral information. Notably, a new multi-branch deep fused feature algorithm based on ResNet (DFFResNet) is introduced, which combines Magnetic Resonance Images (MRI) from different sequences to enhance information complementarity and integration. We conducted prediction experiments on a dataset from the Radiology Department of the First Hospital of Lanzhou University, comprising 117 individuals and seven MRI sequences. The model was trained on 80% of the data using 10-fold cross-validation, and the remaining 20% were used for testing. This evaluation was processed in two cases: CROI, containing samples with a complete region of interest (ROI), and PROI, containing samples with a partial ROI region. The robustness results from repeated experiments at both image and patient levels demonstrate the superior performance and improved generalization of the proposed method compared to alternative models. Our approach yields highly competitive prediction results even when the ROI region outline is incomplete, offering a novel and effective multi-sequence fused strategy for predicting preoperative MVI.

PMID:39196745 | DOI:10.1109/JBHI.2024.3451331

IEEE Trans Neural Syst Rehabil Eng. 2024 Aug 28;PP. doi: 10.1109/TNSRE.2024.3451010. Online ahead of print.

ABSTRACT

We propose EEG-SimpleConv, a straightforward 1D convolutional neural network for Motor Imagery decoding in BCI. Our main motivation is to propose a simple and performing baseline that achieves high classification accuracy, using only standard ingredients from the literature, to serve as a standard for comparison. The proposed architecture is composed of standard layers, including 1D convolutions, batch normalisations, ReLU activation functions and pooling functions. EEG-SimpleConv architecture is accompanied by a straightforward and tailored training routine, which is subjected to an extensive ablation study to quantify the influence of its components. We evaluate its performance on four EEG Motor Imagery datasets, including simulated online setups, and compare it to recent Deep Learning and Machine Learning approaches. EEG-SimpleConv is at least as good or far more efficient than other approaches, showing strong knowledge-transfer capabilities across subjects, at the cost of a low inference time. We believe that using standard components and ingredients can significantly help the adoption of Deep Learning approaches for BCI. We make the code of the models and the experiments accessible.

PMID:39196743 | DOI:10.1109/TNSRE.2024.3451010

Protein Sci. 2024 Jul;33(7):e5076. doi: 10.1002/pro.5076.

ABSTRACT

Antimicrobial resistance is a critical public health concern, necessitating the exploration of alternative treatments. While antimicrobial peptides (AMPs) show promise, assessing their toxicity using traditional wet lab methods is both time-consuming and costly. We introduce tAMPer, a novel multi-modal deep learning model designed to predict peptide toxicity by integrating the underlying amino acid sequence composition and the three-dimensional structure of peptides. tAMPer adopts a graph-based representation for peptides, encoding ColabFold-predicted structures, where nodes represent amino acids and edges represent spatial interactions. Structural features are extracted using graph neural networks, and recurrent neural networks capture sequential dependencies. tAMPer's performance was assessed on a publicly available protein toxicity benchmark and an AMP hemolysis data we generated. On the latter, tAMPer achieves an F1-score of 68.7%, outperforming the second-best method by 23.4%. On the protein benchmark, tAMPer exhibited an improvement of over 3.0% in the F1-score compared to current state-of-the-art methods. We anticipate tAMPer to accelerate AMP discovery and development by reducing the reliance on laborious toxicity screening experiments.

PMID:39196703 | DOI:10.1002/pro.5076

Transl Vis Sci Technol. 2024 Aug 1;13(8):43. doi: 10.1167/tvst.13.8.43.

ABSTRACT

PURPOSE: This study aims to investigate the prevalence of artifacts in optical coherence tomography (OCT) images with acceptable signal strength and evaluate the performance of supervised deep learning models in improving OCT image quality assessment.

METHODS: We conducted a retrospective study on 4555 OCT images from 546 patients, with each image having an acceptable signal strength (≥6). A comprehensive analysis of prevalent OCT artifacts was performed, and five pretrained convolutional neural network models were trained and tested to infer images based on quality.

RESULTS: Our results showed a high prevalence of artifacts in OCT images with acceptable signal strength. Approximately 21% of images were labeled as nonacceptable quality. The EfficientNetV2 model demonstrated superior performance in classifying OCT image quality, achieving an area under the receiver operating characteristic curve of 0.950 ± 0.007 and an area under the precision recall curve of 0.985 ± 0.002.

CONCLUSIONS: The findings highlight the limitations of relying solely on signal strength for OCT image quality assessment and the potential of deep learning models in accurately classifying image quality.

TRANSLATIONAL RELEVANCE: Application of the deep learning-based OCT image quality assessment models may improve the OCT image data quality for both clinical applications and research.

PMID:39196579 | DOI:10.1167/tvst.13.8.43

Int J Cardiovasc Imaging. 2024 Aug 28. doi: 10.1007/s10554-024-03222-8. Online ahead of print.

ABSTRACT

To preliminarily verify the feasibility of a deep-learning (DL) artificial intelligence (AI) model to localize pulmonary embolism (PE) on unenhanced chest-CT by comparison with pulmonary artery (PA) CT angiography (CTA). In a monocentric study, we retrospectively reviewed 99 oncological patients (median age in years: 64 (range: 28-92 years); percentage of female: 39.4%) who received unenhanced and contrast-enhanced chest CT examinations in one session between January 2020 and October 2022 and who were diagnosed incidentally with PE. Findings in the unenhanced images were correlated with the contrast-enhanced images, which were considered the gold standard for central, segmental and subsegmental PE. The new algorithm was trained and tested based on the 99 unenhanced chest-CT image data sets. Based on them, candidate boxes, which were output by the model, were post-processed by evaluating whether the predicted box intersects with the patient's lung segmentation at any position. The AI-based algorithm proved to have an overall sensitivity of 54.5% for central, of 81.9% for segmental and 80.0% for subsegmental PE if taking n = 20 candidate boxes into account. Depending on the localization of the pulmonary embolism, the detection rate for only one box was: 18.1% central, 34.7% segmental and 0.0% subsegmental. The median volume of the clots differed significantly between the three subgroups and was 846.5 mm3 (IQR:591.1-964.8) in central, 201.3 mm3 (IQR:98.3-390.9) in segmental and 110.6 mm3 (IQR:94.3-128.0) in subsegmental PA (p < 0.05). The new algorithm proved to have high sensitivity in detecting PE in particular in segmental/subsegmental localization and may guide to decide whether a second contrast enhanced CT is necessary.

PMID:39196450 | DOI:10.1007/s10554-024-03222-8

Nanoscale. 2024 Aug 28. doi: 10.1039/d4nr01667j. Online ahead of print.

ABSTRACT

Optical multilayer thin films have a wide range of applications due to their ability to manipulate transmissive or reflective wavelengths by adjusting the thickness of composed layers, enabling diverse uses. Although their light weight, flexible nature and ease of fabrication position them as promising components for future devices, determining their optimal layer thickness for the desired functionality demands extensive simulations, leading to inefficient utilization of computational resources and time. To overcome these challenges, inverse design methods, leveraging machine learning and deep learning, are being explored. However, these methods necessitate learning processes, despite the presence of well-established formulas that elucidate these phenomena. Furthermore, deriving accurate answers for conditions not included in the learning process proves to be challenging. This paper introduces an innovative inverse design approach that utilizes the backpropagation of a networked transfer matrix, effectively explaining the characteristics of optical multilayer thin films. By exploiting the chain rule of the network, this method calculates gradients to discern how each layer thickness influences the outcomes. Consequently, the optimal thickness is determined without the need for an additional learning process. Mathematical elucidation of the operational principle of this approach is precisely described. Optimization of computing resource utilization through network configuration reduces the calculation time compared to conventional methods. The efficacy of this method is demonstrated through its application in the inverse design of transmissive and reflective films, verifying its potential for enhancing efficiency and accuracy in optical multilayer thin-film design and manufacturing processes.

PMID:39196333 | DOI:10.1039/d4nr01667j

Environ Sci Pollut Res Int. 2024 Aug 28. doi: 10.1007/s11356-024-34750-4. Online ahead of print.

ABSTRACT

Water resource management relies heavily on reliable water quality predictions. Predicting water quality metrics in the watershed system, including dissolved oxygen (DO), is the main emphasis of this work. The enhanced long short-term memory (LSTM) model was suggested to improve the model's performance. Additionally, a hybrid model was employed to calculate the ideal parameter values for the LSTM model, which helped overcome the nonstationarity, unpredictability, and nonlinearity of the data about the water quality parameters. This model recruited the COOT method. The original weekly water quality values at the Vaigai River, Madurai, Tamil Nadu, India, were tested using the suggested hybrid model. An independent LSTM, the hybrid optimisation method takes its cues from the cuckoo bird's reproductive strategy and a novel meta-heuristic optimisation technique dubbed COOT, which is based on the behaviour of a flock of coot birds. If implemented, the suggested hybrid model might serve as an alternate framework for water quality prediction, laying the groundwork for basin-wide efforts to manage water quality and control pollutants.

PMID:39196324 | DOI:10.1007/s11356-024-34750-4

Int J Numer Method Biomed Eng. 2024 Aug 28:e3858. doi: 10.1002/cnm.3858. Online ahead of print.

ABSTRACT

Experimental blood flow measurement techniques are invaluable for a better understanding of cardiovascular disease formation, progression, and treatment. One of the emerging methods is time-resolved three-dimensional phase-contrast magnetic resonance imaging (4D flow MRI), which enables noninvasive time-dependent velocity measurements within large vessels. However, several limitations hinder the usability of 4D flow MRI and other experimental methods for quantitative hemodynamics analysis. These mainly include measurement noise, corrupt or missing data, low spatiotemporal resolution, and other artifacts. Traditional filtering is routinely applied for denoising experimental blood flow data without any detailed discussion on why it is preferred over other methods. In this study, filtering is compared to different singular value decomposition (SVD)-based machine learning and autoencoder-type deep learning methods for denoising and filling in missing data (imputation). An artificially corrupted and voxelized computational fluid dynamics (CFD) simulation as well as in vitro 4D flow MRI data are used to test the methods. SVD-based algorithms achieve excellent results for the idealized case but severely struggle when applied to in vitro data. The autoencoders are shown to be versatile and applicable to all investigated cases. For denoising, the in vitro 4D flow MRI data, the denoising autoencoder (DAE), and the Noise2Noise (N2N) autoencoder produced better reconstructions than filtering both qualitatively and quantitatively. Deep learning methods such as N2N can result in noise-free velocity fields even though they did not use clean data during training. This work presents one of the first comprehensive assessments and comparisons of various classical and modern machine-learning methods for enhancing corrupt cardiovascular flow data in diseased arteries for both synthetic and experimental test cases.

PMID:39196308 | DOI:10.1002/cnm.3858

Nat Cardiovasc Res. 2022 Apr;1(4):286-288. doi: 10.1038/s44161-022-00044-6.

NO ABSTRACT

PMID:39196129 | DOI:10.1038/s44161-022-00044-6

Nat Cardiovasc Res. 2022 May;1(5):532. doi: 10.1038/s44161-022-00075-z.

NO ABSTRACT

PMID:39195951 | DOI:10.1038/s44161-022-00075-z

Toxins (Basel). 2024 Aug 13;16(8):354. doi: 10.3390/toxins16080354.

ABSTRACT

Fusarium head blight (FHB) is a plant disease caused by various species of the Fusarium fungus. One of the major concerns associated with Fusarium spp. is their ability to produce mycotoxins. Mycotoxin contamination in small grain cereals is a risk to human and animal health and leads to major economic losses. A reliable site-specific precise Fusarium spp. infection early warning model is, therefore, needed to ensure food and feed safety by the early detection of contamination hotspots, enabling effective and efficient fungicide applications, and providing FHB prevention management advice. Such precision farming techniques contribute to environmentally friendly production and sustainable agriculture. This study developed a predictive model, Sága, for on-site FHB detection in wheat using imaging spectroscopy and deep learning. Data were collected from an experimental field in 2021 including (1) an experimental field inoculated with Fusarium spp. (52.5 m × 3 m) and (2) a control field (52.5 m × 3 m) not inoculated with Fusarium spp. and sprayed with fungicides. Imaging spectroscopy data (hyperspectral images) were collected from both the experimental and control fields with the ground truth of Fusarium-infected ear and healthy ear, respectively. Deep learning approaches (pretrained YOLOv5 and DeepMAC on Global Wheat Head Detection (GWHD) dataset) were used to segment wheat ears and XGBoost was used to analyze the hyperspectral information related to the wheat ears and make predictions of Fusarium-infected wheat ear and healthy wheat ear. The results showed that deep learning methods can automatically detect and segment the ears of wheat by applying pretrained models. The predictive model can accurately detect infected areas in a wheat field, achieving mean accuracy and F1 scores exceeding 89%. The proposed model, Sága, could facilitate the early detection of Fusarium spp. to increase the fungicide use efficiency and limit mycotoxin contamination.

PMID:39195764 | DOI:10.3390/toxins16080354

J Chem Theory Comput. 2024 Aug 28. doi: 10.1021/acs.jctc.4c00824. Online ahead of print.

ABSTRACT

The local structure and thermophysical properties of SrCl2-KCl-MgCl2 melt were revealed by deep potential molecular dynamicsdriven by machine learning to facilitate the development of molten salt electrolytic Mg-Sr alloys. The short- and intermediate-range order of the SrCl2-KCl-MgCl2 melts was explored through radial distribution functions and structure factors, respectively, and their component and temperature dependence were discussed comprehensively. In the MgCl2-rich system, the intermediate-range order is more pronounced, and its evolution with temperature exhibits a non-Debye-Waller behavior. Mg-Cl is dominated by 4,5 coordination and Sr-Cl by 6,7 coordination, and their coordination geometries exhibit distorted octahedra and distorted pentagonal bipyramids, respectively. A database of thermophysical properties of SrCl2-KCl-MgCl2 melts, including density, self-diffusion coefficient, viscosity, and ionic conductivity, was thus developed, covering the temperature range from 873 to 1173 K.

PMID:39195736 | DOI:10.1021/acs.jctc.4c00824

Tomography. 2024 Aug 13;10(8):1263-1276. doi: 10.3390/tomography10080094.

ABSTRACT

Anterior cruciate ligament (ACL) tears are prevalent knee injures, particularly among active individuals. Accurate and timely diagnosis is essential for determining the optimal treatment strategy and assessing patient prognosis. Various previous studies have demonstrated the successful application of deep learning techniques in the field of medical image analysis. This study aimed to develop a deep learning model for detecting ACL tears in knee magnetic resonance Imaging (MRI) to enhance diagnostic accuracy and efficiency. The proposed model consists of three main modules: a Dual-Scale Data Augmentation module (DDA) to enrich the training data on both the spatial and layer scales; a selective group attention module (SG) to capture relationships across the layer, channel, and space scales; and a fusion module to explore the inter-relationships among various perspectives to achieve the final classification. To ensure a fair comparison, the study utilized a public dataset from MRNet, comprising knee MRI scans from 1250 exams, with a focus on three distinct views: axial, coronal, and sagittal. The experimental results demonstrate the superior performance of the proposed model, termed SGNET, in ACL tear detection compared with other comparison models, achieving an accuracy of 0.9250, a sensitivity of 0.9259, a specificity of 0.9242, and an AUC of 0.9747.

PMID:39195729 | DOI:10.3390/tomography10080094

Yonsei Med J. 2024 Sep;65(9):527-533. doi: 10.3349/ymj.2023.0590.

ABSTRACT

PURPOSE: This study aimed to develop and validate a convolutional neural network (CNN) that automatically detects an aberrant right subclavian artery (ARSA) on preoperative computed tomography (CT) for thyroid cancer evaluation.

MATERIALS AND METHODS: A total of 556 CT with ARSA and 312 CT with normal aortic arch from one institution were used as the training set for model development. A deep learning model for the classification of patch images for ARSA was developed using two-dimension CNN from EfficientNet. The diagnostic performance of our model was evaluated using external test sets (112 and 126 CT) from two institutions. The performance of the model was compared with that of radiologists for detecting ARSA using an independent dataset of 1683 consecutive neck CT.

RESULTS: The performance of the model was achieved using two external datasets with an area under the curve of 0.97 and 0.99, and accuracy of 97% and 99%, respectively. In the temporal validation set, which included a total of 20 patients with ARSA and 1663 patients without ARSA, radiologists overlooked 13 ARSA cases. In contrast, the CNN model successfully detected all the 20 patients with ARSA.

CONCLUSION: We developed a CNN-based deep learning model that detects ARSA using CT. Our model showed high performance in the multicenter validation.

PMID:39193761 | DOI:10.3349/ymj.2023.0590

J Chem Inf Model. 2024 Aug 28. doi: 10.1021/acs.jcim.4c01020. Online ahead of print.

ABSTRACT

The convergence of biotechnology and artificial intelligence has the potential to transform drug development, especially in the field of therapeutic peptide design. Peptides are short chains of amino acids with diverse therapeutic applications that offer several advantages over small molecular drugs, such as targeted therapy and minimal side effects. However, limited oral bioavailability and enzymatic degradation have limited their effectiveness. With advances in deep learning techniques, innovative approaches to peptide design have become possible. In this work, we demonstrate HYDRA, a hybrid deep learning approach that leverages the distribution modeling capabilities of a diffusion model and combines it with a binding affinity maximization algorithm that can be used for de novo design of peptide binders for various target receptors. As an application, we have used our approach to design therapeutic peptides targeting proteins expressed by Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) genes. The ability of HYDRA to generate peptides conditioned on the target receptor's binding sites makes it a promising approach for developing effective therapies for malaria and other diseases.

PMID:39193724 | DOI:10.1021/acs.jcim.4c01020

Circ Arrhythm Electrophysiol. 2024 Aug 28:e012959. doi: 10.1161/CIRCEP.124.012959. Online ahead of print.

NO ABSTRACT

PMID:39193715 | DOI:10.1161/CIRCEP.124.012959

Comput Cardiol (2010). 2023 Oct;50. doi: 10.22489/cinc.2023.047. Epub 2023 Dec 26.

ABSTRACT

The 12-lead electrocardiogram (ECG) is the most common front-line diagnosis tool for assessing cardiovascular health, yet traditional ECG analysis cannot detect many diseases. Machine learning (ML) techniques have emerged as a powerful set of techniques for producing automated and robust ECG analysis tools that can often predict diseases and conditions not detectable by traditional ECG analysis. Many contemporary ECG-ML studies have focused on utilizing the full 12-lead ECG; however, with the increased availability of single-lead ECG data from wearable devices, there is a clear motivation to explore the development of single-lead ECG-ML techniques. In this study we developed and applied a deep learning architecture for the detection of low left ventricular ejection fraction (LVEF), and compared the performance of this architecture when it was trained with individual leads of the 12-lead ECG to the performance when trained using the entire 12-lead ECG. We observed that single-lead-trained networks performed similarly to the full 12-lead-trained network. We also noted patterns of agreement and disagreement between network low LVEF predictions across the different lead-trained networks.

PMID:39193485 | PMC:PMC11349306 | DOI:10.22489/cinc.2023.047

Digit Health. 2024 Aug 25;10:20552076241269555. doi: 10.1177/20552076241269555. eCollection 2024 Jan-Dec.

ABSTRACT

OBJECTIVE: This study describes a research protocol for a behavioral marker-based predictive model that examines the functional status of older adults with subjective cognitive decline and mild cognitive impairment.

METHODS: A total of 130 older adults aged ≥65 years with subjective cognitive decline or mild cognitive impairment will be recruited from the Dementia Relief Centers or the Community Service Centers. Data on behavioral and psychosocial markers (e.g. physical activity, mobility, sleep/wake patterns, social interaction, and mild behavioral impairment) will be collected using passive wearable actigraphy, in-person questionnaires, and smartphone-based ecological momentary assessments. Two follow-up assessments will be performed at 12 and 24 months after baseline. Mixed-effect machine learning models: MErf, MEgbm, MEmod, and MEctree, and standard machine learning models without random effects [random forest, gradient boosting machine] will be employed in our analyses to predict functional status over time.

RESULTS: The results of this study will be fundamental for developing tailored digital interventions that apply deep learning techniques to behavioral data to predict, identify, and aid in the management of functional decline in older adults with subjective cognitive decline and mild cognitive impairment. These older adults are considered the optimal target population for preventive interventions and will benefit from such tailored strategies.

CONCLUSIONS: Our study will contribute to the development of self-care interventions that utilize behavioral data and machine learning techniques to provide automated analyses of the functional decline of older adults who are at risk for dementia.

PMID:39193313 | PMC:PMC11348489 | DOI:10.1177/20552076241269555

Digit Health. 2024 Aug 20;10:20552076241271639. doi: 10.1177/20552076241271639. eCollection 2024 Jan-Dec.

ABSTRACT

OBJECTIVE: The presence of a lightweight convolutional neural network (CNN) model with a high-accuracy rate and low complexity can be useful in building an early obesity detection system, especially on mobile-based applications. The previous works of the CNN model for obesity detection were focused on the accuracy performances without considering the complexity size. In this study, we aim to build a new lightweight CNN model that can accurately classify normal and obese thermograms with low complexity sizes.

METHODS: The DenseNet201 CNN architectures were modified by replacing the standard convolution layers with multiple depthwise and pointwise convolution layers from the MobileNet architectures. Then, the depth network of the dense block was reduced to determine which depths were the most comparable to obtain minimum validation losses. The proposed model then was compared with state-of-the-art DenseNet and MobileNet CNN models in terms of classification performances, and complexity size, which is measured in model size and computation cost.

RESULTS: The results of the testing experiment show that the proposed model has achieved an accuracy of 81.54% with a model size of 1.44 megabyte (MB). This accuracy was comparable to that of DenseNet, which was 83.08%. However, DenseNet's model size was 71.77 MB. On the other hand, the proposed model's accuracy was higher than that of MobileNetV2, which was 79.23%, with a computation cost of 0.69 billion floating-point operations per second (GFLOPS), which approximated that of MobileNetV2, which was 0.59 GFLOPS.

CONCLUSIONS: The proposed model inherited the feature-extracting ability from the DenseNet201 architecture while keeping the lightweight complexity characteristic of the MobileNet architecture.

PMID:39193310 | PMC:PMC11348482 | DOI:10.1177/20552076241271639