Pub Date : 2025-12-01Epub Date: 2025-06-13DOI: 10.1016/j.mex.2025.103440
Bipin Nair B J , Niranjan , Saketh P , Shobha Rani N
Handwritten Braille character recognition presents a significant challenge in the field of assistive technology, especially with the inclusion of various linguistic scripts such as Kannada. The data set is uniquely curated, combining ground-truth data from Kaggle and real-world samples collected from blind schools, segmented into vowels and consonants. The proposed system demonstrates exceptional performance in feature extraction, classification accuracy, and addressing spatial misalignments in Braille dots. Comparative analysis against state-of-the-art methods confirms the efficiency of the proposed model in overcoming the limitations of conventional techniques. The system was trained with two train test splits 70:30 and 80:20. The initial train test split has achieved 97.9 % and the latter one has achieved 98.7 %. This study aims to contribute significantly to the empowerment of visually impaired communities through advancements in automated Braille recognition systems.
•
The study addresses the challenge of handwritten Kannada Braille recognition using a uniquely curated dataset from Kaggle and blind schools, divided into vowels and consonants.
•
The proposed system achieves high accuracy (97.9 % for 70:30 and 98.7 % for 80:20 split) showing superior feature extraction and handling of spatial misalignments in Braille dots.
•
Comparative analysis of state-of-the-art methods confirms the model’s efficiency in overcoming limitations of conventional techniques, contributing to assistive technology for visually impaired communities.
{"title":"WAYVision: A hybrid deep learning approach for recognizing handwritten Kannada Braille using wavelet transformation and attention based YOLOv5","authors":"Bipin Nair B J , Niranjan , Saketh P , Shobha Rani N","doi":"10.1016/j.mex.2025.103440","DOIUrl":"10.1016/j.mex.2025.103440","url":null,"abstract":"<div><div>Handwritten Braille character recognition presents a significant challenge in the field of assistive technology, especially with the inclusion of various linguistic scripts such as Kannada. The data set is uniquely curated, combining ground-truth data from Kaggle and real-world samples collected from blind schools, segmented into vowels and consonants. The proposed system demonstrates exceptional performance in feature extraction, classification accuracy, and addressing spatial misalignments in Braille dots. Comparative analysis against state-of-the-art methods confirms the efficiency of the proposed model in overcoming the limitations of conventional techniques. The system was trained with two train test splits 70:30 and 80:20. The initial train test split has achieved 97.9 % and the latter one has achieved 98.7 %. This study aims to contribute significantly to the empowerment of visually impaired communities through advancements in automated Braille recognition systems.<ul><li><span>•</span><span><div>The study addresses the challenge of handwritten Kannada Braille recognition using a uniquely curated dataset from Kaggle and blind schools, divided into vowels and consonants.</div></span></li><li><span>•</span><span><div>The proposed system achieves high accuracy (97.9 % for 70:30 and 98.7 % for 80:20 split) showing superior feature extraction and handling of spatial misalignments in Braille dots.</div></span></li><li><span>•</span><span><div>Comparative analysis of state-of-the-art methods confirms the model’s efficiency in overcoming limitations of conventional techniques, contributing to assistive technology for visually impaired communities.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103440"},"PeriodicalIF":1.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144501159","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-08-16DOI: 10.1016/j.mex.2025.103564
Anusha R, Srinivas Prasad
Cervical cancer is a serious health concern that entails high risks for individuals due to delayed detection and treatment worldwide. Formal screening for the condition is challenging in developing countries due to several factors, including medical costs, access to healthcare facilities, and delayed symptom manifestation. A blockchain-enabled healthcare system for cervical cancer risk prediction ensures data security, privacy, and accurate risk assessment. This system uses blockchain to provide decentralised, tamper-proof storage and access control over sensitive patient data, ensuring that only authorized entities can interact with the information. An improved spotted hyena optimization algorithm is employed for cervical cancer risk prediction, fine-tuning a Graph Convolutional Network (GCN) integrated with an Attention Mechanism and a Gated Recurrent Unit (GRU). The GCN captures complex relationships between medical attributes and patients, while the attention mechanism dynamically assigns weights to features based on relevance, improving predictive accuracy. The GRU processes sequential data, such as medical history, to model temporal dependencies in the risk factors. The metaheuristic optimization further enhances the model by finding the optimal parameters, boosting performance
Introduces a blockchain-enabled system for secure and decentralized medical data management
Applies an intelligent model for predicting cervical cancer risk using patient health records
Demonstrates improved accuracy, privacy, and reliability over traditional diagnostic methods
{"title":"A blockchain-enabled healthcare system for cervical cancer risk prediction using enhanced metaheuristic optimised graph convolutional attention based GRU","authors":"Anusha R, Srinivas Prasad","doi":"10.1016/j.mex.2025.103564","DOIUrl":"10.1016/j.mex.2025.103564","url":null,"abstract":"<div><div>Cervical cancer is a serious health concern that entails high risks for individuals due to delayed detection and treatment worldwide. Formal screening for the condition is challenging in developing countries due to several factors, including medical costs, access to healthcare facilities, and delayed symptom manifestation. A blockchain-enabled healthcare system for cervical cancer risk prediction ensures data security, privacy, and accurate risk assessment. This system uses blockchain to provide decentralised, tamper-proof storage and access control over sensitive patient data, ensuring that only authorized entities can interact with the information. An improved spotted hyena optimization algorithm is employed for cervical cancer risk prediction, fine-tuning a Graph Convolutional Network (GCN) integrated with an Attention Mechanism and a Gated Recurrent Unit (GRU). The GCN captures complex relationships between medical attributes and patients, while the attention mechanism dynamically assigns weights to features based on relevance, improving predictive accuracy. The GRU processes sequential data, such as medical history, to model temporal dependencies in the risk factors. The metaheuristic optimization further enhances the model by finding the optimal parameters, boosting performance</div><div>Introduces a blockchain-enabled system for secure and decentralized medical data management</div><div>Applies an intelligent model for predicting cervical cancer risk using patient health records</div><div>Demonstrates improved accuracy, privacy, and reliability over traditional diagnostic methods</div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103564"},"PeriodicalIF":1.9,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144896174","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
To expedite the early stages of drug development for diseases lacking established target databases, and to enhance knowledge updating in well-studied disease domains, this paper introduces TARGETFLOW, an automated literature-mining pipeline. The workflow begins by automatically retrieving literature, downloading relevant abstracts, and constructing a comprehensive database. After performing selective text cleaning and data preprocessing, it leverages large language models (LLMs) to conduct intelligent literature screening, followed by code-based whitespace tokenization. Subsequently, rule-based filtering is applied to extract high-potential therapeutic targets for the specified disease. To validate the effectiveness of this pipeline, three hypotheses were formulated: (1) An effective pipeline should be capable of identifying high-potential therapeutic targets for the given disease; (2) For diseases with established target databases, the pipeline should be able to detect novel and emerging targets not yet included in existing databases; and (3) The pipeline should also be applicable to rare or emerging diseases that lack mature target databases. Then, rheumatoid arthritis (RA), a common disease, and idiopathic pulmonary fibrosis (IPF), a rare disease, were selected as case studies. The results demonstrated the method’s reliability (high-potential target validation rate: 56 %), innovativeness (new target validation pass rate: 100 %), and generalizability (IPF target literature support rate: 88.9 %).
{"title":"TARGETFLOW:An automated literature mining pipeline for accelerating the discovery of high-potential targets in rare and mature diseases","authors":"Shuhan Guo, Tianxiang Shang, Yuzhu Pan, Yingjia Wu, Yan Li, Mohan Zhou","doi":"10.1016/j.mex.2025.103735","DOIUrl":"10.1016/j.mex.2025.103735","url":null,"abstract":"<div><div>To expedite the early stages of drug development for diseases lacking established target databases, and to enhance knowledge updating in well-studied disease domains, this paper introduces TARGETFLOW, an automated literature-mining pipeline. The workflow begins by automatically retrieving literature, downloading relevant abstracts, and constructing a comprehensive database. After performing selective text cleaning and data preprocessing, it leverages large language models (LLMs) to conduct intelligent literature screening, followed by code-based whitespace tokenization. Subsequently, rule-based filtering is applied to extract high-potential therapeutic targets for the specified disease. To validate the effectiveness of this pipeline, three hypotheses were formulated: (1) An effective pipeline should be capable of identifying high-potential therapeutic targets for the given disease; (2) For diseases with established target databases, the pipeline should be able to detect novel and emerging targets not yet included in existing databases; and (3) The pipeline should also be applicable to rare or emerging diseases that lack mature target databases. Then, rheumatoid arthritis (RA), a common disease, and idiopathic pulmonary fibrosis (IPF), a rare disease, were selected as case studies. The results demonstrated the method’s reliability (high-potential target validation rate: 56 %), innovativeness (new target validation pass rate: 100 %), and generalizability (IPF target literature support rate: 88.9 %).</div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103735"},"PeriodicalIF":1.9,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145681155","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-06-27DOI: 10.1016/j.mex.2025.103474
Yihan Zhang, Shanshan Li
This study proposes a logistic regression-integrated cellular automata (CA) model for oil spill simulation, addressing challenges in parameter determination of traditional CA models. The method involves data preprocessing (geospatial alignment, resampling, normalization), Monte Carlo sampling for training data, logistic regression-based weight assignment to impact factors, neighborhood function and stochastic term computation, and iterative oil spill simulation. The model can be calibrated through sensitivity analyses of sampling ratios, spatial scales, and neighborhood structures. Finally, it was validated using DeepSpill experimental data. Results show optimal accuracy (97.40 %) under 22 % sampling ratio, 12.61 % oil area proportion, 6 m spatial scale, and 7 × 7 Moore neighborhood.
•
Innovative Model Integration & Calibration: Merged logistic regression with CA to objectively quantify environmental drivers (currents, wind, salinity) and optimize parameters (sampling, scale and neighborhood) in oil simulation.
•
Dynamic Optimization & Scale Sensitivity: Peak accuracy (96.41 %) can be obtained at 22 % sampling rate and 12.61 % oil area. 97.32 % accuracy at 6 m resolution balances resolution and boundary roughness.
{"title":"A method to modelling oil spill using combination of logistic regression and cellular automata","authors":"Yihan Zhang, Shanshan Li","doi":"10.1016/j.mex.2025.103474","DOIUrl":"10.1016/j.mex.2025.103474","url":null,"abstract":"<div><div>This study proposes a logistic regression-integrated cellular automata (CA) model for oil spill simulation, addressing challenges in parameter determination of traditional CA models. The method involves data preprocessing (geospatial alignment, resampling, normalization), Monte Carlo sampling for training data, logistic regression-based weight assignment to impact factors, neighborhood function and stochastic term computation, and iterative oil spill simulation. The model can be calibrated through sensitivity analyses of sampling ratios, spatial scales, and neighborhood structures. Finally, it was validated using DeepSpill experimental data. Results show optimal accuracy (97.40 %) under 22 % sampling ratio, 12.61 % oil area proportion, 6 m spatial scale, and 7 × 7 Moore neighborhood.<ul><li><span>•</span><span><div>Innovative Model Integration & Calibration: Merged logistic regression with CA to objectively quantify environmental drivers (currents, wind, salinity) and optimize parameters (sampling, scale and neighborhood) in oil simulation.</div></span></li><li><span>•</span><span><div>Dynamic Optimization & Scale Sensitivity: Peak accuracy (96.41 %) can be obtained at 22 % sampling rate and 12.61 % oil area. 97.32 % accuracy at 6 m resolution balances resolution and boundary roughness.</div></span></li><li><span>•</span><span><div>Neighborhood-Driven Diffusion Enhancement: 7 × 7 Moore neighborhood boosts accuracy to 97.40 % (vs. 3 × 3), proving neighborhood size critically shapes diffusion dynamics.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103474"},"PeriodicalIF":1.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144570202","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-09-30DOI: 10.1016/j.mex.2025.103652
N. Shobha Rani , Bhavya K R , I. Jeena Jacob , Pushpa B. R , Bipin Nair BJ , Akshatha Prabhu
The reliable classification of medicinal plant species plays a vital role in ensuring their quality, authenticity, and safe use in healthcare. However, existing methods often face difficulties when species exhibit strong visual similarities or when datasets are imbalanced, which limits their effectiveness in practice. Although deep learning models such as ResNet18 and VGG16 have proven influential in image recognition tasks, our experiments showed that they tended to overfit, with validation losses reaching 42.99 % and test accuracy falling to 73.99 % in certain groups. To overcome these challenges, we introduce a multi-level fusion feature model that combines 3D normalized color histograms, extended uniform Local Binary Patterns (LBP with P = 24, R = 3), multi-orientation Gabor filters, and Histogram of Oriented Gradients (HOG). This approach captures a richer set of visual cues by bringing together global color statistics, detailed textures, frequency-domain patterns, and shape descriptors. We incorporate SMOTE-based synthetic augmentation to address further class imbalance, which helps balance feature distributions across categories. We employ a soft-voting ensemble of machine learning classifiers for classification and use cosine similarity metrics to capture inter-class relationships better. Tests on Indian medicinal plant datasets show that our model consistently outperforms deep learning baselines, reaching 100 % accuracy in Group 1, 95.82 % in Group 3, and over 90 % in other groups. These results suggest that the proposed model offers a more robust and computationally efficient solution for plant species classification, particularly under conditions of high inter-class similarity and dataset imbalance.
•
The proposed domain-specific model can be applied explicitly to Indian plant species groups exhibiting high inter-class visual similarities through a novel feature fusion strategy.
•
The proposed multi-level feature fusion method's innovation integrates 3D normalized color histograms, extended uniform LBP (P = 24, R = 3), multi-orientation Gabor filters, and HOG features to capture the color, texture, and shape characteristics.
•
The proposed work offers a scalable ensemble framework for inter-class similarity analysis by combining SMOTE-based class balancing, feature normalization, and a soft-voting ensemble of diverse classifiers that support biodiversity and ecological studies.
药用植物物种的可靠分类对保证其质量、真实性和在医疗保健中的安全使用起着至关重要的作用。然而，当物种表现出强烈的视觉相似性或当数据集不平衡时，现有的方法往往面临困难，这限制了它们在实践中的有效性。尽管ResNet18和VGG16等深度学习模型已被证明在图像识别任务中具有影响力，但我们的实验表明，它们倾向于过拟合，在某些组中验证损失达到42.99%，测试准确率下降到73.99%。为了克服这些挑战，我们引入了一种多层次融合特征模型，该模型结合了3D归一化颜色直方图、扩展均匀局部二值模式（LBP, P = 24, R = 3）、多向Gabor滤波器和定向梯度直方图（HOG）。这种方法通过将全局颜色统计、详细纹理、频域模式和形状描述符结合在一起，捕获了一组更丰富的视觉线索。我们结合了基于smote的合成增强来解决进一步的类别不平衡，这有助于平衡类别之间的特征分布。我们采用机器学习分类器的软投票集成进行分类，并使用余弦相似度度量来更好地捕获类间关系。对印度药用植物数据集的测试表明，我们的模型始终优于深度学习基线，在第1组中达到100%的准确率，在第3组中达到95.82%，在其他组中达到90%以上。这些结果表明，该模型为植物物种分类提供了一种鲁棒性和计算效率更高的解决方案，特别是在类间相似性高和数据不平衡的情况下。•提出的领域特定模型可以通过一种新颖的特征融合策略明确应用于表现出高度类间视觉相似性的印度植物物种群。•提出的多层次特征融合方法的创新之处是集成了3D归一化颜色直方图、扩展均匀LBP （P = 24, R = 3）、多向Gabor滤波器和HOG特征来捕获颜色、纹理和形状特征。•提出的工作提供了一个可扩展的集成框架，通过结合基于smote的类平衡、特征规范化和支持生物多样性和生态研究的各种分类器的软投票集成，用于类间相似性分析。
{"title":"CISCS: Classification of inter-class similarity based medicinal plant species groups with machine learning","authors":"N. Shobha Rani , Bhavya K R , I. Jeena Jacob , Pushpa B. R , Bipin Nair BJ , Akshatha Prabhu","doi":"10.1016/j.mex.2025.103652","DOIUrl":"10.1016/j.mex.2025.103652","url":null,"abstract":"<div><div>The reliable classification of medicinal plant species plays a vital role in ensuring their quality, authenticity, and safe use in healthcare. However, existing methods often face difficulties when species exhibit strong visual similarities or when datasets are imbalanced, which limits their effectiveness in practice. Although deep learning models such as ResNet18 and VGG16 have proven influential in image recognition tasks, our experiments showed that they tended to overfit, with validation losses reaching 42.99 % and test accuracy falling to 73.99 % in certain groups. To overcome these challenges, we introduce a multi-level fusion feature model that combines 3D normalized color histograms, extended uniform Local Binary Patterns (LBP with <em>P</em> = 24, <em>R</em> = 3), multi-orientation Gabor filters, and Histogram of Oriented Gradients (HOG). This approach captures a richer set of visual cues by bringing together global color statistics, detailed textures, frequency-domain patterns, and shape descriptors. We incorporate SMOTE-based synthetic augmentation to address further class imbalance, which helps balance feature distributions across categories. We employ a soft-voting ensemble of machine learning classifiers for classification and use cosine similarity metrics to capture inter-class relationships better. Tests on Indian medicinal plant datasets show that our model consistently outperforms deep learning baselines, reaching 100 % accuracy in Group 1, 95.82 % in Group 3, and over 90 % in other groups. These results suggest that the proposed model offers a more robust and computationally efficient solution for plant species classification, particularly under conditions of high inter-class similarity and dataset imbalance.<ul><li><span>•</span><span><div>The proposed domain-specific model can be applied explicitly to Indian plant species groups exhibiting high inter-class visual similarities through a novel feature fusion strategy.</div></span></li><li><span>•</span><span><div>The proposed multi-level feature fusion method's innovation integrates 3D normalized color histograms, extended uniform LBP (<em>P</em> = 24, <em>R</em> = 3), multi-orientation Gabor filters, and HOG features to capture the color, texture, and shape characteristics.</div></span></li><li><span>•</span><span><div>The proposed work offers a scalable ensemble framework for inter-class similarity analysis by combining SMOTE-based class balancing, feature normalization, and a soft-voting ensemble of diverse classifiers that support biodiversity and ecological studies.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103652"},"PeriodicalIF":1.9,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145320096","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-07-17DOI: 10.1016/j.mex.2025.103519
Abhay Nath , Om Roy , Priyanka Silveri , Sanskruti Patel
Oral squamous cell carcinoma (OSCC) continues to present a major worldwide healthcare problem because patients have poor survival outcomes alongside frequent disease returns. Globocan predicts that, OSCC will result in 389,846 new cases and 188,438 deaths globally during 2022 while maintaining an extremely poor 5-year survival rate at about 50%. Our method applies residual connections with Squeeze-and-Excitation blocks along with hybrid attention systems and enhanced activation functions and optimization algorithms to boost gradient movement throughout feature extraction. Compared against established conventional CNN backbones (VGG16, ResNet50, DenseNet121, and more), the proposed ConvNeXt-SE-Attn model outperformed them in all aspects of discrimination and calibration, including precision 97.88% (vs. ≤94.2%), sensitivity 96.82% (vs. ≤92.5%), specificity 95.94% (vs. ≤93.1%), F1 score 97.31% (vs. ≤93.8%), AUC 0.9644 (vs. ≤0.945), and MCC 0.9397 (vs. ≤0.910). The findings are critical to the increased feature-representation power and the robustness of classification of the architecture.
The proposed architecture employs ConvNeXt backbone with SE blocks and hybrid attention to extract essential details within class boundaries which standard models usually miss.
The activation through Gaussian-based GReLU incorporates Swish activation together with DropPath regularization for producing smooth gradient patterns which lead to generalizable features across imbalanced datasets.
Grad-CAM enhances interpretability by showing which image sections lead to predictions in order to enable clinical decisions.
The model demonstrates its capability as an effective detection method for minimal variations in oral cells which supports precise non-invasive treatment approaches for OSCC.
{"title":"Deep learning approach with ConvNeXt-SE-attn model for in vitro oral squamous cell carcinoma and chemotherapy analysis","authors":"Abhay Nath , Om Roy , Priyanka Silveri , Sanskruti Patel","doi":"10.1016/j.mex.2025.103519","DOIUrl":"10.1016/j.mex.2025.103519","url":null,"abstract":"<div><div>Oral squamous cell carcinoma (OSCC) continues to present a major worldwide healthcare problem because patients have poor survival outcomes alongside frequent disease returns. Globocan predicts that, OSCC will result in 389,846 new cases and 188,438 deaths globally during 2022 while maintaining an extremely poor 5-year survival rate at about 50%. Our method applies residual connections with Squeeze-and-Excitation blocks along with hybrid attention systems and enhanced activation functions and optimization algorithms to boost gradient movement throughout feature extraction. Compared against established conventional CNN backbones (VGG16, ResNet50, DenseNet121, and more), the proposed ConvNeXt-SE-Attn model outperformed them in all aspects of discrimination and calibration, including precision 97.88% (vs. ≤94.2%), sensitivity 96.82% (vs. ≤92.5%), specificity 95.94% (vs. ≤93.1%), F1 score 97.31% (vs. ≤93.8%), AUC 0.9644 (vs. ≤0.945), and MCC 0.9397 (vs. ≤0.910). The findings are critical to the increased feature-representation power and the robustness of classification of the architecture.</div><div>The proposed architecture employs ConvNeXt backbone with SE blocks and hybrid attention to extract essential details within class boundaries which standard models usually miss.</div><div>The activation through Gaussian-based GReLU incorporates Swish activation together with DropPath regularization for producing smooth gradient patterns which lead to generalizable features across imbalanced datasets.</div><div>Grad-CAM enhances interpretability by showing which image sections lead to predictions in order to enable clinical decisions.</div><div>The model demonstrates its capability as an effective detection method for minimal variations in oral cells which supports precise non-invasive treatment approaches for OSCC.</div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103519"},"PeriodicalIF":1.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144696758","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-07-23DOI: 10.1016/j.mex.2025.103530
Rogerio Cichota, Xiumei Yang
The methodology used to develop and test a crop system model to simulate growth and defoliation management of plantain forage (Plantago lanceolata L.) in the Agricultural Production Systems Simulator (APSIM) framework is presented. The model has been primarily developed based on data from New Zealand, but it should be applicable in other regions, and it can be extended to account for new cultivars being released in the market. The model is capable of simulating pure forage plantain stands and can also be used in mixed swards with other forage species within the APSIM framework. Validation tests for the current stage focused on pure forage plantain stands in New Zealand; results showed a good model performance when predicting biomass accumulation over a growing season and nitrogen content (with NSE > 0.8), although predictions for individual defoliation events were not very accurate (NSE between -1.0 and 0.6). More data is needed to improve how the model describes biomass partition among plant organs and how this is affected by environmental conditions and defoliations. This model contributes to the ongoing efforts to explore alternative crops and improve management of grazing systems, aiming to reduce environmental impacts while maintaining productivity.
•
Data to describe various aspects of plantain forage were gathered from literature and field trials
•
Model built to simulate plant growth and re-growth after defoliations within APSIM
•
The model can be used in monoculture and mixed swards
{"title":"Simulating forage plantain growth and defoliation in APSIM","authors":"Rogerio Cichota, Xiumei Yang","doi":"10.1016/j.mex.2025.103530","DOIUrl":"10.1016/j.mex.2025.103530","url":null,"abstract":"<div><div>The methodology used to develop and test a crop system model to simulate growth and defoliation management of plantain forage (<em>Plantago lanceolata</em> L.) in the Agricultural Production Systems Simulator (APSIM) framework is presented. The model has been primarily developed based on data from New Zealand, but it should be applicable in other regions, and it can be extended to account for new cultivars being released in the market. The model is capable of simulating pure forage plantain stands and can also be used in mixed swards with other forage species within the APSIM framework. Validation tests for the current stage focused on pure forage plantain stands in New Zealand; results showed a good model performance when predicting biomass accumulation over a growing season and nitrogen content (with NSE > 0.8), although predictions for individual defoliation events were not very accurate (NSE between -1.0 and 0.6). More data is needed to improve how the model describes biomass partition among plant organs and how this is affected by environmental conditions and defoliations. This model contributes to the ongoing efforts to explore alternative crops and improve management of grazing systems, aiming to reduce environmental impacts while maintaining productivity.<ul><li><span>•</span><span><div>Data to describe various aspects of plantain forage were gathered from literature and field trials</div></span></li><li><span>•</span><span><div>Model built to simulate plant growth and re-growth after defoliations within APSIM</div></span></li><li><span>•</span><span><div>The model can be used in monoculture and mixed swards</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103530"},"PeriodicalIF":1.9,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144721624","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-06-13DOI: 10.1016/j.mex.2025.103431
Adrián Domínguez-Diaz , Manuel Goyanes , Luis de-Marcos
This paper presents a protocol for using ChatGPT to perform content analysis. The protocol involves converting a codebook, outlining categories, descriptions, coding rules, and possible values, into a structured prompt that guides ChatGPT's analysis. The protocol was validated through analysis of 980 research articles to identify research approaches and data collection methods. ChatGPT achieved high performance in identifying data collection methods, but faced challenges with poorly defined or underrepresented categories, particularly in mixed methods research. Overall, while it scored well for quantitative (0.96) and qualitative (0.82) studies, it struggled with mixed methods (0.60), highlighting the need for clear methodological definitions.
•
The protocol enhances coding efficiency and demonstrates the feasibility of using AI for content analysis, potentially streamlining the coding process in research.
•
Challenges arose in categories that were not clearly defined (big data), underrepresented (ethnography), or hierarchically related (Interview & Discourse/Textual analysis).
•
Interrater metrics indicated a substantial level of agreement, reinforcing the potential of ChatGPT in content analysis while emphasizing the importance of clear methodological definitions.
{"title":"Automating content analysis of scientific abstracts using ChatGPT: A methodological protocol and use case","authors":"Adrián Domínguez-Diaz , Manuel Goyanes , Luis de-Marcos","doi":"10.1016/j.mex.2025.103431","DOIUrl":"10.1016/j.mex.2025.103431","url":null,"abstract":"<div><div>This paper presents a protocol for using ChatGPT to perform content analysis. The protocol involves converting a codebook, outlining categories, descriptions, coding rules, and possible values, into a structured prompt that guides ChatGPT's analysis. The protocol was validated through analysis of 980 research articles to identify research approaches and data collection methods. ChatGPT achieved high performance in identifying data collection methods, but faced challenges with poorly defined or underrepresented categories, particularly in mixed methods research. Overall, while it scored well for quantitative (0.96) and qualitative (0.82) studies, it struggled with mixed methods (0.60), highlighting the need for clear methodological definitions.<ul><li><span>•</span><span><div>The protocol enhances coding efficiency and demonstrates the feasibility of using AI for content analysis, potentially streamlining the coding process in research.</div></span></li><li><span>•</span><span><div>Challenges arose in categories that were not clearly defined (big data), underrepresented (ethnography), or hierarchically related (Interview & Discourse/Textual analysis).</div></span></li><li><span>•</span><span><div>Interrater metrics indicated a substantial level of agreement, reinforcing the potential of ChatGPT in content analysis while emphasizing the importance of clear methodological definitions.</div></span></li></ul></div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103431"},"PeriodicalIF":1.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144321571","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-10-28DOI: 10.1016/j.mex.2025.103688
Jose Manuel Valencia-Moreno, Everardo Gutierrez-Lopez, Jose Angel Gonzalez-Fraga, Rodolfo Alan Martinez Rodriguez, Olivia Denisse Victoria Mejia, Alma Alejandra Soberano Serrano
This article presents a dataset of breast cancer risk factors collected from 1697 Cuban women between 2001 and 2018, as a tool to design and support the development and validation of predictive models in public health for breast cancer risk. A reproducible methodology for quality control and variable enrichment was implemented to ensure data integrity and compatibility with machine learning techniques.
• Reproducible preprocessing methodology to ensure data quality and traceability.
• Open breast cancer risk factor dataset for epidemiological studies and risk assessment using machine learning.
• Consistent prediction model performance across multiple metrics after data preprocessing
{"title":"Data preparation method for machine learning-based breast cancer risk prediction: A Cuban case study","authors":"Jose Manuel Valencia-Moreno, Everardo Gutierrez-Lopez, Jose Angel Gonzalez-Fraga, Rodolfo Alan Martinez Rodriguez, Olivia Denisse Victoria Mejia, Alma Alejandra Soberano Serrano","doi":"10.1016/j.mex.2025.103688","DOIUrl":"10.1016/j.mex.2025.103688","url":null,"abstract":"<div><div>This article presents a dataset of breast cancer risk factors collected from 1697 Cuban women between 2001 and 2018, as a tool to design and support the development and validation of predictive models in public health for breast cancer risk. A reproducible methodology for quality control and variable enrichment was implemented to ensure data integrity and compatibility with machine learning techniques.</div><div>• Reproducible preprocessing methodology to ensure data quality and traceability.</div><div>• Open breast cancer risk factor dataset for epidemiological studies and risk assessment using machine learning.</div><div>• Consistent prediction model performance across multiple metrics after data preprocessing</div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103688"},"PeriodicalIF":1.9,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145465861","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Missing data in climate time series is a significant problem because it complicates the monitoring and prediction of climatic phenomena. The primary objective of this research document is to describe the most relevant imputation methods for missing data in the climate context over the last decade. Results reveal a superior concentration of documents on the use of imputation methods for climate time series in Asia and Europe, with notable examples from Malaysia, China, and Italy. Meanwhile, Brazil and Australia were the countries with a high number of research in America and Oceania. Moreover, temperature and precipitation were the most frequently employed climate variables. Regarding the information source, the monitoring networks were the most commonly used source for extracting data in almost all the research. On the other hand, methods such as mean techniques, simple and multiple linear regression, interpolation, and Principal Component Analysis (PCA) were the conventional statistical techniques used for imputing missing data. Furthermore, artificial neural networks demonstrated the ability to identify complex patterns in the data. Finally, Generative Adversarial Networks excel over other deep learning methods in the imputation of missing climate data.
{"title":"Missing data imputation of climate time series: A review","authors":"Lizette Elena Alejo-Sanchez , Aldo Márquez-Grajales , Fernando Salas-Martínez , Anilu Franco-Arcega , Virgilio López-Morales , Otilio Arturo Acevedo-Sandoval , César Abelardo González-Ramírez , Ramiro Villegas-Vega","doi":"10.1016/j.mex.2025.103455","DOIUrl":"10.1016/j.mex.2025.103455","url":null,"abstract":"<div><div>Missing data in climate time series is a significant problem because it complicates the monitoring and prediction of climatic phenomena. The primary objective of this research document is to describe the most relevant imputation methods for missing data in the climate context over the last decade. Results reveal a superior concentration of documents on the use of imputation methods for climate time series in Asia and Europe, with notable examples from Malaysia, China, and Italy. Meanwhile, Brazil and Australia were the countries with a high number of research in America and Oceania. Moreover, temperature and precipitation were the most frequently employed climate variables. Regarding the information source, the monitoring networks were the most commonly used source for extracting data in almost all the research. On the other hand, methods such as mean techniques, simple and multiple linear regression, interpolation, and Principal Component Analysis (PCA) were the conventional statistical techniques used for imputing missing data. Furthermore, artificial neural networks demonstrated the ability to identify complex patterns in the data. Finally, Generative Adversarial Networks excel over other deep learning methods in the imputation of missing climate data.</div></div>","PeriodicalId":18446,"journal":{"name":"MethodsX","volume":"15 ","pages":"Article 103455"},"PeriodicalIF":1.6,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144489601","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号