Handling class imbalance in education using data-level and deep learning methods

Acute Lymphoblastic Leukemia (ALL) is the most common pediatric cancer, requiring early detection to save lives and reduce the financial burden of advanced-stage treatment. While traditional diagnostic methods are time-consuming and resource-intensive, deep transfer learning offers a computationally efficient alternative for medical image classification. This study employed two widely recognized transfer learning algorithms, VGG-19 and EfficientNet-B3, to detect ALL using a publicly available dataset of 10,661 images from 118 patients. Data preprocessing included resizing, augmentation, and normalization. The models were trained for 100 epochs, with batch sizes of 30 for VGG-19 and 32 for EfficientNet-B3. Evaluation metrics such as accuracy, precision, recall, and F1 score were used to assess model performance. Statistical significance testing was performed using paired t-tests (p<0.05). Comparative analysis was performed with existing studies to validate the findings. EfficientNet-B3 significantly outperformed VGG-19, achieving an average accuracy of 96% compared to 80% for VGG-19 (p<0.001). EfficientNet-B3 demonstrated superior performance in handling class imbalance, with the minority class (Hem) achieving precision, recall, and F1 scores of 97%, 89%, and 93%, respectively. VGG-19 struggled with the minority class, achieving lower recall (51%) and F1 score (62%). However, dataset limitations including single-source origin may affect generalizability. This study highlights the effectiveness of EfficientNet-B3 as a reliable tool for early ALL detection, offering high accuracy and computational efficiency. Clinical implementation requires addressing computational constraints and integration challenges. Future research could integrate multimodal datasets to identify risk factors and further improve diagnostic accuracy.

Objective:Social and behavioral determinants of health play a critical role in patient outcomes, yet much of this information is documented only in unstructured clinical text rather than structured records. Detecting these factors using natural language processing enables a deeper understanding of patient health and supports better decision-making. This study aims to improve the prediction of Behavioral Determinants of Health (BDoH) from medical records by systematically comparing multiple transformer-based models: Bio-ClinicalBERT, BioBERT, BioMedBERT, RoBERTa, and T5-EHR with and without BiLSTM. We also address the challenge of class imbalance in the dataset and assess the role of generative versus discriminative models in this task.Methods:We evaluated five transformer models in combination with BiLSTM for classifying BDoH mentions in the MIMIC-III dataset. To address the challenge of class imbalance, we compared oversampling, undersampling, and class weighting strategies to identify the most effective approach. Then, each model was first tested as a standalone classifier to establish baseline performance. We then extracted embeddings from these models and used them as input to a BiLSTM layer to investigate whether sequential modeling could further improve classification. Also, by introducing our T5-EHR model alongside BERT and RoBERTa models, we were able to assess both the benefits of BiLSTM and the value of using a generative model for this task. Finally, we benchmarked the performance of the best approaches against the published results from the MIMIC-SBDH study, using precision, recall, and F1 scores as evaluation metrics.Results:The experiments showed that class weighting was the most effective strategy for handling class imbalance, showing better performance across all models. The addition of BiLSTM improved the performance of Bio-ClinicalBERT, BioMedBERT, and BioBERT across all categories. RoBERTa and T5-EHR achieved strong results as standalone models, with no additional benefit from BiLSTM integration. Across all evaluated labels, T5-EHR achieved the highest F1 scores compared to the other models and prior baselines in this study.Conclusion:This study demonstrates that handling class imbalance is essential for robust prediction of social and behavioral determinants of health from clinical notes. We find that the integration of BiLSTM provides clear benefits for Bio-ClinicalBERT and BioBERT, enhancing their ability to capture sequential dependencies. However, RoBERTa and T5-EHR achieved their best performance without BiLSTM, reflecting the strength of their pretrained representations. Among all models, our pretrained T5-EHR, adapted especially for clinical text, outperformed every baseline, establishing the highest F1 scores across all labels and surpassing the results reported in the MIMIC-SBDH benchmark. These findings highlight both the value of model-specific strategies and the effectiveness of generative models for advancing state-of-the-art classification in health informatics.

Use of Machine Learning to Identify Follow-Up Recommendations in Radiology Reports

It is crucial to realize the municipal solid waste (MSW) classification in terms of its treatments and disposals. Deep learning used for the classification of residual waste and wet waste from MSW was considered as a promising method. While few studies reported using the method of deep learning with transfer learning to classify organic waste and residual waste. Thus, this study aims to discuss the effect of the transfer learning on the performance of different deep learning structures, VGGNet-16 and ResNet-50, for the classification of organic waste and residual waste, which were compared in terms of the training time, confusion matric, accuracy, precision, and recall. In addition, the algorithms of PCA and t-SNE were also adopted to compare the representation extracted from the last layer of various deep learning models. Results indicated that transfer learning could shorten the training time and the training time of various deep learning follows this order: VGGNet-16 (402 s) &amp;gt; VGGNet-16 with TL (272 s) &amp;gt; ResNet-50 (238 s) &amp;gt; ResNet-50 with TL (223 s). Compared with the method of PAC, waste representations were better separated from high dimension to low dimension by t-SNE. The values of organic waste in terms of F1 score follows this order: ResNet-50 with transfer learning (97.8%) &amp;gt; VGGNet-16 with transfer learning (97.1%) &amp;gt; VGGNet-16 (95.0%) &amp;gt; ResNet-50 (92.5%).Therefore, the best performance for the classification of organic and residual waste was ResNet-50 with transfer learning, followed by VGGNet-16 with transfer learning and VGGNet-16, and ResNet-50 in terms of accuracy, precision, recall, and F1 score.

Farmland is the cornerstone of agriculture and is important for food security and social production. Farmland assessment is essential but traditional methods are usually expensive and slow. Deep learning methods have been developed and widely applied recently in image recognition, semantic understanding, and many other application domains. In this research, we used fully convolutional networks (FCN) as the deep learning model to evaluate farmland grades. Normalized difference vegetation index (NDVI) derived from Landsat images was used as the input data, and the China National Cultivated Land Grade Database within Jiangsu Province was used to train the model on cloud computing. We also applied an image segmentation method to improve the original results from the FCN and compared the results with classical machine learning (ML) methods. Our research found that the FCN can predict farmland grades with an overall F1 score (the harmonic mean of precision and recall) of 0.719 and F1 score of 0.909, 0.590, 0.740, 0.642, and 0.023 for non-farmland, level I, II, III, and IV farmland, respectively. Combining the FCN and image segmentation method can further improve prediction accuracy with results of fewer noise pixels and more realistic edges. Compared with conventional ML, at least in farmland evaluation, FCN provides better results with higher precision, recall, and F1 score. Our research indicates that by using remote sensing NDVI data, the deep learning method can provide acceptable farmland assessment without fieldwork and can be used as a novel supplement to traditional methods. The method used in this research will save a lot of time and cost compared with traditional means.

AI-generated content (AIGC) in the context of dermoscopic image analysis describes the application of artificial intelligence (AI) approaches to produce synthetic images for training and enriching machine learning (ML) methods. In dermatology, where dermoscopic images are essential in skin lesion detection, AIGC helps overcome the challenges of inadequate datasets and class imbalances. By employing advanced generative complex systems like Generative Adversarial Networks (GANs), AIGC will simulate different dermoscopic conditions by generating realistic-view synthetic images. These AI-generated images help to enhance the training dataset, contributing to the ML models with many widespread and varied sets of instances. This aids in increasing the ability of the model to precisely predict and generalize while handling new, unnoticed dermoscopic images. Deep learning (DL) methods, specifically convolutional neural networks (CNNs), can provide a massive breakthrough in an extensive range of computer vision (CV) tasks of complex systems, predominantly by exploiting large-scale annotated datasets. This paper develops a GAN-based Synthetic Medical Image Augmentation for the imbalanced Dermoscopic Image Analysis (GANSMIA-CDIA) method. The GANSMIA-CDIA technique’s primary target is exploiting the GAN model for synthetic image generation to handle class imbalance data problems. In addition, the GANSMIA-CDIA technique diagnoses the melanoma using the optimal DL model. The GANSMIA-CDIA technique applies a contrast enhancement process for the noise eradication process. The GANSMIA-CDIA technique follows a feature fusion method comprising different DL methods such as MobileNetV2, AlexNet, ResNet50, and Inceptionv3 to learn feature patterns in the preprocessed images. Meanwhile, the Social Network Search (SNS) technique is utilized for the hyperparameter tuning process. Also, the bidirectional long short-term memory (BiLSTM) technique is implemented to detect and classify melanoma. A series of simulations were performed on the standard dermoscopic image dataset to evaluate the performance of the GANSMIA-CDIA technique. The experimental values indicate the excellence of the GANSMIA-CDIA technique over existing techniques.

Now a day’s emotion prediction model plays an important role in computer communication system. Deep learning is emerging area in prediction of various tasks. Proposed method is used to predict human emotions using deep learning. Convolutional neural networks (CNNs) play an emerging role in identifying emotions from human expressions such as sad, happy, surprise, angry, neutral and fear. Proposed model employs FER 2013 and CK+ open source available datasets for training and testing to find emotion. Proposed research compared deep learning model with machine learning algorithms and achieved 95% accuracy is higher than existing methods. Model performance obtained from other performance metrics such as recall, precision and F1 score and accuracy shows better than existing methods. This research demonstrates how deep learning methods have the potential to transform emotion recognition software and make a significant contribution to the field of affective AI.

Deep learning methods have become popular among researchers in the field of fault detection. However, their performance depends on the availability of big datasets. To overcome this problem researchers started applying transfer learning to achieve good performance from small available datasets, by leveraging multiple prediction models over similar machines and working conditions. However, the influence of negative transfer limits their application. Negative transfer among prediction models increases when the environment and working conditions are changing continuously. To overcome the effect of negative transfer, we propose a novel deep transfer learning method, coined deep boosted transfer learning, for wind turbine gearbox fault detection that prevents negative transfer and only focuses on relevant information from the source machine. The proposed method is an instance-based deep transfer learning method that updates the weights of the source and the target machine training samples separately. The weights of different source training samples are gradually decreased to reduce the impact on the final model. The proposed method is verified by the Case Western Reserve University bearing and real field wind farm datasets. The results show that the proposed method ignores negative transfer and achieves higher accuracy compared to standard deep learning and deep transfer learning methods.

Every year, lung cancer contributes to a high percentage deaths in the world. Early detection of lung cancer is important for its effective treatment, and non-invasive rapid methods are usually used for diagnosis. In this study, we aimed to detect lung cancer using deep learning methods and determine the contribution of deep learning to the classification of lung carcinoma using a convolutional neural network (CNN). A total of 301 patients diagnosed with lung carcinoma pathologies in our hospital were included in the study. In the thorax, Computed Tomography (CT) was performed for diagnostic purposes prior to the treatment. After tagging the section images, tumor detection, small and non-small cell lung carcinoma differentiation, adenocarcinoma-squamous cell lung carcinoma differentiation, and adenocarcinoma-squamous cell-small cell lung carcinoma differentiation were sequentially performed using deep CNN methods. In total, 301 lung carcinoma images were used to detect tumors, and the model obtained with the deep CNN system exhibited 0.93 sensitivity, 0.82 precision, and 0.87 F1 score in detecting lung carcinoma. In the differentiation of small cell-non-small cell lung carcinoma, the sensitivity, precision and F1 score of the CNN model at the test stage were 0.92, 0.65, and 0.76, respectively. In the adenocarcinoma-squamous cancer differentiation, the sensitivity, precision, and F1 score were 0.95, 0.80, and 0.86, respectively. The patients were finally grouped as small cell lung carcinoma, adenocarcinoma, and squamous cell lung carcinoma, and the CNN model was used to determine whether it could differentiate these groups. The sensitivity, specificity, and F1 score of this model were 0.90, 0.44, and 0.59, respectively, in this differentiation. In this study, we successfully detected tumors and differentiated between adenocarcinoma- squamous cell carcinoma groups with the deep learning method using the CNN model. Due to their non-invasive nature and the success of the deep learning methods, they should be integrated into radiology to diagnose lung carcinoma.

The class imbalance problem is the main problem in classification. This issue arises because real-world datasets frequently exhibit an imbalance as a result of a class with more instances than other classes. In handling class imbalance, a Hybrid Approach that blends data-level and algorithm-level approaches produce good results. However, apart from the class imbalance, which reduces classification accuracy, the complexity of the data also has an effect. The complexity of this data causes a minority noise sample which lies between the minority and the majority. In order to determine how close minority samples are to their homogeneous and heterogeneous nearest neighbors, it is necessary to calculate the relative density. The greater the proximity to the homogeneous nearest neighbors, the greater the relative density, which causes the minority samples to be in a safe state but otherwise be categorized as noisy samples. This research will combine the application of the Hybrid Approach with A self-adaptive Robust SMOTE (RSMOTE), which is an adaptive method from SMOTE that applies the concept of relative density in the over-sampling process on minority samples. The research contribution is to implement the Hybrid Approach-RSMOTE in handling class imbalance with noise by using relative density in over-sampling and also to improve classification performance. The results showed that the Hybrid Approach-RSMOTE and Hybrid Approach-SMOTE had given good results in handling class imbalance. However, the Hybrid Approach-RSMOTE gave better results in the Precision, Recall, F1-Measure, and G-Mean and showed significant differences. Based on the results of the study, it can be stated that the performance of the Hybrid Approach in handling class imbalance is influenced by the selection of the over-sampling method. The results show that RSMOTE can be considered an over-sampling method in the Hybrid Approach.

To assess the performance of a deep learning method for detecting the segmentation of periapical lesions on dental panoramic radiographs. Observational study. Place and Duration of the Study: Faculty of Dentistry, Van Yuzuncu YilUniversity, Van, Turkiye, from Marchto September 2024. The deep learning model, YOLOv5, based on the YOLO algorithm for periapical lesion segmentation, was further developed using 1,500 anonymised panoramic radiographs. The radiographs were obtained from the Radiology Archive at the aforementioned University. For apical lesion segmentation, YOLOv5 with the PyTorch model was utilised. The dataset was divided into training (n = 1,200 radiographs / 2,628 labels), validation (150 radiographs / 325 labels), and test (n = 150 radiographs / 368 labels) sets. The model's effectiveness was measured using the confusion matrix. Sensitivity (recall), precision, and F1 scores provided quantitative assessments of the model's predictive capabilities. The sensitivity, precision, and F1 score performance values of the YOLOv5 deep learning algorithm were 0.682, 0.784, and 0.729, respectively. Periapical lesions on panoramic radiography can be reliably identified using deep learning algorithms. Dental healthcare is being revolutionised by artificial intelligence and deep learning methods, which are advantageous to both the system and practitioners. While the current YOLO-based system yields encouraging findings, additional data should be gathered in future research to improve detection outcomes. Panoramic radiography, Periapical pathology, Deep learning, Artificial intelligence, Lesion segmentation, YOLOv5.

Educational data mining helps the educational institutions to perform effectively and efficiently by exploiting the data related to all its stakeholders. It can help the at-risk students, develop recommendation systems and alert the students at different levels. It is beneficial to the students, educators and authorities as a whole. Deep learning has gained momentum in various domains especially image processing with a large dataset. We devise a regression model for analyzing the academic performance of the students using deep learning. We have applied regression using deep learning and linear regression on the dataset. For such models with smaller datasets, to tackle the issue of overfitting is critical. Hence, the parameters can be tuned to deal with such issues. The deep learning model records a mean absolute score (mae) of 1.61 and loss 4.7 with the value of k = 3. While the linear regression model yields a loss of 6.7 and mae score of 1.97. The deep learning model outperforms the linear regression model. The model may be successfully extended to other programmes to mine and predict the performance of the learners.

Less than 10% of patients with cancer have detectable pathogenic germline alterations, which may be partially due to incomplete pathogenic variant detection. To evaluate if deep learning approaches identify more germline pathogenic variants in patients with cancer. A cross-sectional study of a standard germline detection method and a deep learning method in 2 convenience cohorts with prostate cancer and melanoma enrolled in the US and Europe between 2010 and 2017. The final date of clinical data collection was December 2017. Germline variant detection using standard or deep learning methods. The primary outcomes included pathogenic variant detection performance in 118 cancer-predisposition genes estimated as sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). The secondary outcomes were pathogenic variant detection performance in 59 genes deemed actionable by the American College of Medical Genetics and Genomics (ACMG) and 5197 clinically relevant mendelian genes. True sensitivity and true specificity could not be calculated due to lack of a criterion reference standard, but were estimated as the proportion of true-positive variants and true-negative variants, respectively, identified by each method in a reference variant set that consisted of all variants judged to be valid from either approach. The prostate cancer cohort included 1072 men (mean [SD] age at diagnosis, 63.7 [7.9] years; 857 [79.9%] with European ancestry) and the melanoma cohort included 1295 patients (mean [SD] age at diagnosis, 59.8 [15.6] years; 488 [37.7%] women; 1060 [81.9%] with European ancestry). The deep learning method identified more patients with pathogenic variants in cancer-predisposition genes than the standard method (prostate cancer: 198 vs 182; melanoma: 93 vs 74); sensitivity (prostate cancer: 94.7% vs 87.1% [difference, 7.6%; 95% CI, 2.2% to 13.1%]; melanoma: 74.4% vs 59.2% [difference, 15.2%; 95% CI, 3.7% to 26.7%]), specificity (prostate cancer: 64.0% vs 36.0% [difference, 28.0%; 95% CI, 1.4% to 54.6%]; melanoma: 63.4% vs 36.6% [difference, 26.8%; 95% CI, 17.6% to 35.9%]), PPV (prostate cancer: 95.7% vs 91.9% [difference, 3.8%; 95% CI, -1.0% to 8.4%]; melanoma: 54.4% vs 35.4% [difference, 19.0%; 95% CI, 9.1% to 28.9%]), and NPV (prostate cancer: 59.3% vs 25.0% [difference, 34.3%; 95% CI, 10.9% to 57.6%]; melanoma: 80.8% vs 60.5% [difference, 20.3%; 95% CI, 10.0% to 30.7%]). For the ACMG genes, the sensitivity of the 2 methods was not significantly different in the prostate cancer cohort (94.9% vs 90.6% [difference, 4.3%; 95% CI, -2.3% to 10.9%]), but the deep learning method had a higher sensitivity in the melanoma cohort (71.6% vs 53.7% [difference, 17.9%; 95% CI, 1.82% to 34.0%]). The deep learning method had higher sensitivity in the mendelian genes (prostate cancer: 99.7% vs 95.1% [difference, 4.6%; 95% CI, 3.0% to 6.3%]; melanoma: 91.7% vs 86.2% [difference, 5.5%; 95% CI, 2.2% to 8.8%]). Among a convenience sample of 2 independent cohorts of patients with prostate cancer and melanoma, germline genetic testing using deep learning, compared with the current standard genetic testing method, was associated with higher sensitivity and specificity for detection of pathogenic variants. Further research is needed to understand the relevance of these findings with regard to clinical outcomes.

Educational Data Mining (EDM) uncovers insights from large datasets collected from various educational platforms, such as online learning systems, student information databases, and classroom tools. EDM helps educators identify hidden patterns that improve teaching strategies, personalize learning experiences, and predict student performance. Predicting student success has become a key focus of EDM, allowing institutions to implement targeted interventions and personalized support. The dataset included academic achievement grades from 1,001 students enrolled in various courses during the fall semester across multiple years, to demonstrate how proposed models provide more accurate predictions compared to traditional machine learning methods. Models such as YOLO, Fast R-CNN, Artificial Neural Networks (ANNs), and Long Short-Term Memory (LSTM) networks are used to capture complex, non-linear relationships within the data. The comparative analysis shows that these deep learning models significantly outperform traditional techniques, such as decision trees and support vector machines (SVMs). The results indicate that proposed method offers improved predictive accuracy, enabling educational institutions to identify at-risk students and deliver tailored interventions. This study highlights the potential of enhanced method to transform personalized education and enhance student success by better understanding individual learning needs and behaviors.

Bacteria, viruses, and fungi impact plant development. So early diseases diagnosis and prevention are crucial for plants. Many different machine-learning approaches are used to identify and classify rice plant illnesses. As technology improves, Deep Learning methods, including DL algorithms, boost precision and accuracy. DL gives segmented-part detail. Early plant disease specialists saw and analyse without a sophisticated system. This takes additional labour and processing time. A physical analysis isn't practical owing to the size of the agricultural industry. However, image processing using Deep Learning may be utilized to identify and classify rice plant diseases. It detects and classifies plant illnesses automatically. The report also covers present successes, limits, and proposals for newborn investigation linked with rice insufficiency disorders. Pre-processing, picture segmentation, feature extraction, and illness classification are discussed. A technique for detecting rice leaf disease based on Deep Convolution Neural Network transfer learning. The method that has been suggested can differentiate between six categories, namely healthy, Rice Tangro, Rice Yellow Mottling, Bacterial Blight, Leaf Streak and Leave Scalding. The upgraded dataset that is not normalized has an accuracy of 97%. The precision, recall, and specificity scores and the F1-score were respectively 97%, 97%, 99% and 97%. The recommended new method performed better than earlier methods that used datasets of the same size or a size comparable to those used in the literature.