Dynamically Balanced Online Random Forests for Interactive Scribble-Based Segmentation

Prediction of body mass index status from voice signals based on machine learning for automated medical applications

Bone age assessment is an important indicator of child maturity. Generally, a person evaluated for bone age is mostly in the puberty stage. Compared to toddlers and post-puberty stages, the data of bone age at the puberty stage are hence much easier to obtain. As a result, the amounts of bone age data collected at the toddlers and post-puberty stages are often less than that of bone age data at the pulpy stage. This causes a data imbalance problem that affects the prediction accuracy. To deal with this problem, a data imbalance immunity bone age assessment (DIIBAA) system is proposed which integrates a CNN-based autoencoder and CNN-based scoring network. The DIIBAA system builds three autoencoders for the bone age data of toddlers, puberty, and post-puberty stages. Since the three types of autoencoders do not interfere with each other, there is no data imbalance problem in the DIIBAA system. After that, the outputs of the three autoencoders are input to the scoring network, and the auto-encoder produced the image with the highest score is regarded as the final prediction result. In the experiments, imbalanced training data with a positive and negative sample ratio of 1:2 were used, which has been alleviated compared to the original very imbalanced data. Meanwhile, the DenseNet169-based autoencoders are employed in the experiments, and the obtained evaluation accuracies are improved.

Bank telemarketing campaigns often experience low subscription rates due to customer heterogeneity and severe class imbalance, which pose challenges for reliable predictive modeling. This study investigates a data-driven approach that integrates synthetic minority oversampling and cost-sensitive learning to improve the prediction of telemarketing outcomes. Experiments are conducted using the Portuguese Bank Marketing dataset, comprising 41,188 instances with a positive response rate of 11.3%. Eight machine learning models are evaluated under a unified preprocessing pipeline and five-fold stratified cross-validation, including Logistic Regression, Decision Tree, Random Forest, and Ensemble methods. The results show that Ensemble models, particularly CatBoost, XGBoost, and LightGBM, achieve improved performance compared with traditional baselines, with notable gains in minority-class recall and overall discrimination ability. The best-performing model attains an F1-score of 0.540, a recall of 0.812 for the positive class, and a ROC–AUC of 0.908. To enhance interpretability, SHAP-based analysis is applied to quantify feature contributions, identifying campaign duration, previous contact outcomes, and selected macroeconomic indicators as key predictors. These findings indicate that combining resampling strategies with cost-sensitive optimization provides a robust and transparent approach for learning from imbalanced telemarketing data, thereby supporting reproducible and data-driven financial decision-making by explicitly addressing difficulty in minority-class identification under imbalance and class imbalance under cross-entropy training in imbalanced banking data.

Random forest (RF) is a competitive machine learning theorem, while one of the big challenges for it is imbalanced real-world data. A Two-dimensional-reduction RF (2DRRF) is presented in this paper, which is optimized based on traditional RF and three innovation points as follows. To improve RF in terms of performance on imbalanced data, a two-dimensional-reduction approach is created. Then, a modified T-link is proposed focusing on detecting and reducing safe samples. Moreover, a biased sampling manner is employed to build up optimal training datasets. Across 13 imbalanced datasets from KEEL-dataset with imbalance-ratio ranging from 6.38 to 129.44, experiments are carried out indicating that 2DRRF steadily holds advantages over the other two relevant implementations of RF in terms of accuracy, recall, precision and F-value.

Interactive segmentation methods utilize user-provided positive and negative clicks to guide the model in accurately segmenting target objects. Compared to fully automatic medical image segmentation, these methods can achieve higher segmentation accuracy with limited image data, demonstrating significant potential in clinical applications. Typically, for each new click provided by the user, conventional interactive segmentation methods reprocess the entire network by re-inputting the click into the segmentation model, which greatly increases the user’s interaction burden and deviates from the intended goal of interactive segmentation tasks. To address this issue, we propose an efficient segmentation network, ESM-Net, for interactive medical image segmentation. It obtains high-quality segmentation masks based on the user’s initial clicks, reducing the complexity of subsequent refinement steps. Recent studies have demonstrated the strong performance of the Mamba model in various vision tasks; however, its application in interactive segmentation remains unexplored. In our study, we incorporate the Mamba module into our framework for the first time and enhance its spatial representation capabilities by developing a Spatial Augmented Convolution (SAC) module. These components are combined as the fundamental building blocks of our network. Furthermore, we designed a novel and efficient segmentation head to fuse multi-scale features extracted from the encoder, optimizing the generation of the predicted segmentation masks. Through comprehensive experiments, our method achieved state-of-the-art performance on three medical image datasets. Specifically, we achieved 1.43 NoC@90 on the Kvasir-SEG dataset, 1.57 NoC@90 on the CVC-ClinicDB polyp segmentation dataset, and 1.03 NoC@90 on the ADAM retinal disk segmentation dataset. The assessments on these three medical image datasets highlight the effectiveness of our approach in interactive medical image segmentation.

PurposeThis study evaluated generative methods to potentially mitigate artificial intelligence (AI) bias when diagnosing diabetic retinopathy (DR) resulting from training data imbalance or domain generalization, which occurs when deep learning systems (DLSs) face concepts at test/inference time they were not initially trained on.MethodsThe public domain Kaggle EyePACS dataset (88,692 fundi and 44,346 individuals, originally diverse for ethnicity) was modified by adding clinician-annotated labels and constructing an artificial scenario of data imbalance and domain generalization by disallowing training (but not testing) exemplars for images of retinas with DR warranting referral (DR-referable) from darker-skin individuals, who presumably have greater concentration of melanin within uveal melanocytes, on average, contributing to retinal image pigmentation. A traditional/baseline diagnostic DLS was compared against new DLSs that would use training data augmented via generative models for debiasing.ResultsAccuracy (95% confidence intervals [CIs]) of the baseline diagnostics DLS for fundus images of lighter-skin individuals was 73.0% (66.9% to 79.2%) versus darker-skin of 60.5% (53.5% to 67.3%), demonstrating bias/disparity (delta = 12.5%; Welch t-test t = 2.670, P = 0.008) in AI performance across protected subpopulations. Using novel generative methods for addressing missing subpopulation training data (DR-referable darker-skin) achieved instead accuracy, for lighter-skin, of 72.0% (65.8% to 78.2%), and for darker-skin, of 71.5% (65.2% to 77.8%), demonstrating closer parity (delta = 0.5%) in accuracy across subpopulations (Welch t-test t = 0.111, P = 0.912).ConclusionsFindings illustrate how data imbalance and domain generalization can lead to disparity of accuracy across subpopulations, and show that novel generative methods of synthetic fundus images may play a role for debiasing AI.Translational RelevanceNew AI methods have possible applications to address potential AI bias in DR diagnostics from fundus pigmentation, and potentially other ophthalmic DLSs too.

Early and accurate breast cancer diagnosis remains critical yet challenging in routine practice. This study proposes a simple, reproducible pipeline that combines deep feature extraction from pre-trained CNNs (ResNet50, VGG16, EfficientNet-B0, DenseNet121, MobileNetV2) with classical machine-learning classifiers (logistic regression, SVM, k-NN, decision tree, random forest, gradient boosting, XGBoost, LightGBM, Naïve Bayes, and MLP). Features are computed after standardized preprocessing; class imbalance is addressed with SMOTE when present. We evaluate three image datasets (binary and multiclass) using accuracy, precision, recall/sensitivity, F1, and confusion matrices, and apply paired statistical tests across cross-validation splits. Findings: EfficientNet-B0+MLP and ResNet50+MLP achieve peak accuracies up to 99.6% on high-quality, balanced data, while DenseNet121+MLP with SMOTE attains 97.8% on imbalanced multiclass data. SMOTE yields substantial gains on imbalanced data and negligible effect on balanced sets; decision trees underperform consistently. Novelty/Improvement: Rather than a monolithic end-to-end network, we provide a modular, resource-aware blueprint that (i) disentangles feature extraction from classification, (ii) quantifies when imbalance correction matters, and (iii) reports clinically relevant error types. We further outline explainability with Grad-CAM/SHAP and discuss inference-time trade-offs and real-world workflow integration, offering an interpretable and deployment-friendly alternative to heavier end-to-end models.

Effectiveness of resampling methods in coping with imbalanced crash data: Crash type analysis and predictive modeling

IntroductionThe prevalence of end-stage renal disease has raised the need for renal replacement therapy over recent decades. Even though a kidney transplant offers an improved quality of life and lower cost of care than dialysis, graft failure is possible after transplantation. Hence, this study aimed to predict the risk of graft failure among post-transplant recipients in Ethiopia using the selected machine learning prediction models.MethodologyThe data was extracted from the retrospective cohort of kidney transplant recipients at the Ethiopian National Kidney Transplantation Center from September 2015 to February 2022. In response to the imbalanced nature of the data, we performed hyperparameter tuning, probability threshold moving, tree-based ensemble learning, stacking ensemble learning, and probability calibrations to improve the prediction results. Merit-based selected probabilistic (logistic regression, naive Bayes, and artificial neural network) and tree-based ensemble (random forest, bagged tree, and stochastic gradient boosting) models were applied. Model comparison was performed in terms of discrimination and calibration performance. The best-performing model was then used to predict the risk of graft failure.ResultsA total of 278 completed cases were analyzed, with 21 graft failures and 3 events per predictor. Of these, 74.8% are male, and 25.2% are female, with a median age of 37. From the comparison of models at the individual level, the bagged tree and random forest have top and equal discrimination performance (AUC-ROC = 0.84). In contrast, the random forest has the best calibration performance (brier score = 0.045). Under testing the individual model as a meta-learner for stacking ensemble learning, the result of stochastic gradient boosting as a meta-learner has the top discrimination (AUC-ROC = 0.88) and calibration (brier score = 0.048) performance. Regarding feature importance, chronic rejection, blood urea nitrogen, number of post-transplant admissions, phosphorus level, acute rejection, and urological complications are the top predictors of graft failure.ConclusionsBagging, boosting, and stacking, with probability calibration, are good choices for clinical risk predictions working on imbalanced data. The data-driven probability threshold is more beneficial than the natural threshold of 0.5 to improve the prediction result from imbalanced data. Integrating various techniques in a systematic framework is a smart strategy to improve prediction results from imbalanced data. It is recommended for clinical experts in kidney transplantation to use the final calibrated model as a decision support system to predict the risk of graft failure for individual patients.

In day today life, data is generated in massive amount with rapid growth in health care environment. The medical industries have large amount of data sets, for diagnosis purpose and maintain patient's records. The medical researches come with new treatments and medicine every day. But availability of medical datasets is often not balanced in their class labels. The performance of some existing method is poor on imbalanced dataset. So the prediction of disease from imbalanced data becomes difficult to handle. In this proposal Classifier ensemble method (Random Forest algorithm) can be used to overcome existing classifier techniques. Multiple classifier system is more accurate and robust than an existing classifier technique. The ensemble method proves to be very efficient in classification of records from available imbalanced healthcare patient data, as it involves the process of considering opinion from multiple base classifiers, as opposed to the single classifier method. This method gives a very accurate and precise inference, as unrelated data's are removed because of multiple base classifiers. The problems of healthcare dataset especially with some uncertainty can be predicted.

Current state-of-the-art natural language processing (NLP) techniques use transformer deep-learning architectures, which depend on large training datasets. We hypothesized that traditional NLP techniques may outperform transformers for smaller radiology report datasets. We compared the performance of BioBERT, a deep-learning-based transformer model pre-trained on biomedical text, and three traditional machine-learning models (gradient boosted tree, random forest, and logistic regression) on seven classification tasks given free-text radiology reports. Tasks included detection of appendicitis, diverticulitis, bowel obstruction, and enteritis/colitis on abdomen/pelvis CT reports, ischemic infarct on brain CT/MRI reports, and medial and lateral meniscus tears on knee MRI reports (7,204 total annotated reports). The performance of NLP models on held-out test sets was compared after training using the full training set, and 2.5%, 10%, 25%, 50%, and 75% random subsets of the training data. In all tested classification tasks, BioBERT performed poorly at smaller training sample sizes compared to non-deep-learning NLP models. Specifically, BioBERT required training on approximately 1,000 reports to perform similarly or better than non-deep-learning models. At around 1,250 to 1,500 training samples, the testing performance for all models began to plateau, where additional training data yielded minimal performance gain. With larger sample sizes, transformer NLP models achieved superior performance in radiology report binary classification tasks. However, with smaller sizes (<1000) and more imbalanced training data, traditional NLP techniques performed better. Our benchmarks can help guide clinical NLP researchers in selecting machine-learning models according to their dataset characteristics.

Class imbalance should not throw you off balance: Choosing the right classifiers and performance metrics for brain decoding with imbalanced data

Effects of class imbalance on resampling and ensemble learning for improved prediction of cyanobacteria blooms

Collaborative filtering based recommendation learns users’ preferences from all users’ historical behavior data, and has been popular to facilitate decision making. Recently, the fairness issue of recommendation has become more and more essential. A recommender system is considered unfair when it does not perform equally well for different user groups according to users’ sensitive attributes (e.g., gender, race). Plenty of methods have been proposed to alleviate unfairness by optimizing a predefined fairness goal or changing the distribution of unbalanced training data. However, they either suffered from the specific fairness optimization metrics or relied on redesigning the current recommendation architecture. In this paper, we study how to improve recommendation fairness from the data augmentation perspective. The recommendation model amplifies the inherent unfairness of imbalanced training data. We augment imbalanced training data towards balanced data distribution to improve fairness. Given each real original user-item interaction record, we propose the following hypotheses for augmenting the training data: each user in one group has a similar item preference (click or non-click) as the item preference of any user in the remaining group. With these hypotheses, we generate “fake" interaction behaviors to complement the original training data. After that, we design a bi-level optimization target, with the inner optimization generates better fake data to augment training data with our hypotheses, and the outer one updates the recommendation model parameters based on the augmented training data. The proposed framework is generally applicable to any embedding-based recommendation, and does not need to pre-define a fairness metric. Extensive experiments on two real-world datasets clearly demonstrate the superiority of our proposed framework. We publish the source code at https://github.com/newlei/FDA.

In this paper, novel cost-sensitive principal component analysis (CSPCA) and cost-sensitive non-negative matrix factorization (CSNMF) methods are proposed for handling the problem of feature extraction from imbalanced data. The presence of highly imbalanced data misleads existing feature extraction techniques to produce biased features, which results in poor classification performance especially for the minor class problem. To solve this problem, we propose a cost-sensitive learning strategy for feature extraction techniques that uses the imbalance ratio of classes to discount the majority samples. This strategy is adapted to the popular feature extraction methods such as PCA and NMF. The main advantage of the proposed methods is that they are able to lessen the inherent bias of the extracted features to the majority class in existing PCA and NMF algorithms. Experiments on twelve public datasets with different levels of imbalance ratios show that the proposed methods outperformed the state-of-the-art methods on multiple classifiers.