Understanding and preventing student dropout presents a decision-critical modeling problem involving heterogeneous variables, nonlinear relationships, and the need for transparent inference. This study addresses the prediction of undergraduate academic outcomes, including Graduation, Enrolled, and Dropout, by proposing a efficientand interpretable machine learning framework that explicitly balances predictive performance, feature efficiency, and algorithmic explainability. The empirical analysis relies on a dataset of 4424 student records across 17 undergraduate programs from the Polytechnic Institute of Portalegre, Portugal. In contrast to existing approaches that rely on high-dimensional input spaces and opaque predictive architectures, we develop a reduced-dimensional classification pipeline based on recursive feature elimination with Gradient Boosting and Random Forest models. Starting from a comprehensive set of demographic, academic, and financial indicators, only 20 informative predictors are retained for model construction, substantially reducing input complexity while preserving predictive capacity. Comparative evaluation across multiple learning algorithms identifies Gradient Boosting as the most effective model, achieving an AUC of 0.891. Beyond predictive accuracy, the proposed framework emphasizes model interpretability through the integration of SHapley Additive exPlanations (SHAP), enabling quantitative attribution of feature contributions at both global and instance levels. The analysis reveals that second-semester academic engagement variables—including the number of courses approved, evaluated, and enrolled—as well as tuition fee payment status and age at enrollment, are the dominant factors shaping student outcomes. Overall, the results demonstrate that strong classification performance can be achieved using a compact feature set while maintaining transparent and explainable model behavior. By combining mathematically grounded feature selection with principled model explanation, this study advances methodological understanding of how efficiency, interpretability, and predictive accuracy can be jointly optimized in applied machine learning, with implications for decision-support systems in educational analytics.

Precise identification of cancer subtypes from whole slide images (WSIs) is pivotal in tailoring patient-specific therapies. Under the weakly supervised multiple instance learning (MIL) paradigm, existing techniques frequently fall short in simultaneously capturing local tissue textures and long-range contextual relationships. To address these challenges, we introduce ConvMixerSSM, a hybrid model that integrates a ConvMixer block for local spatial representation, a state space model (SSM) block for capturing long-range dependencies, and a feature-gated block to enhance informative feature selection. The model was evaluated on the TCGA-NSCLC dataset and the CAMELYON16 dataset for cancer subtyping tasks. Extensive experiments, including comparisons with state-of-the-art MIL methods and ablation studies, were conducted to assess the contribution of each component. ConvMixerSSM achieved an AUC of 97.83%, an ACC of 91.82%, and an F1 score of 91.18%, outperforming existing MIL baselines on the TCGA-NSCLC dataset. The ablation study revealed that each block contributed positively to performance, with the full model showing the most balanced and superior results. Moreover, our visualization results further confirm that ConvMixerSSM can effectively identify tumor regions within WSIs, providing model interpretability and clinical relevance. These findings suggest that ConvMixerSSM has strong potential for advancing computational pathology applications in clinical decision-making.

Sustainable development is multidimensional, yet governance is often examined through aggregate SDG performance, leaving limited understanding of whether different governance dimensions are equally salient across economic, social, and environmental domains. This study investigates domain-specific governance salience using cross-national data from 2015 to 2024. Governance is measured through the Worldwide Governance Indicators, while sustainability performance is modeled through the overall SDG index and three domain-specific indices covering economic, social, and environmental sustainability. Tree-based machine learning and SHAP analysis are used to compare the relative importance of governance indicators and socioeconomic control variables. The findings reveal a domain-sensitive governance pattern that is obscured by aggregate SDG performance. Governance indicators are secondary in the overall SDG model but become more salient in the economic dimension, especially through Government Effectiveness, Rule of Law, and Regulatory Quality. In the social dimension, governance retains moderate relevance alongside health and development-related conditions. By contrast, the environmental domain exhibits limited predictability, highlighting an important boundary condition in the role of governance. Further analysis of high-income countries reveals that specific governance dimensions exhibit greater predictive relevance, although the overall governance pattern remains less consistent. These findings provide support for a metagovernance-informed perspective, showing that the relevance of governance for sustainable development differs across sustainability domains.

Computational pathology (CPath) has significantly advanced the clinical practice of pathology. Despite the progress made, Multiple Instance Learning (MIL), a promising paradigm within CPath, continues to face challenges, particularly related to incomplete information utilization. Existing frameworks, such as those based on Convolutional Neural Networks (CNNs), attention, and selective scan space state sequential model (SSM), lack sufficient flexibility and scalability in fusing diverse features, and cannot effectively fuse diverse features. Additionally, current approaches do not adequately exploit order-related and order-independent features, resulting in suboptimal utilization of sequence information. To address these limitations, we propose a novel MIL framework called Mamba2MIL. Our framework utilizes the state space duality model (SSD) to model long sequences of patches of whole slide images (WSIs), which, combined with weighted feature selection, supports the fusion processing of more branching features and can be extended according to specific application needs. Moreover, we introduce a sequence transformation method tailored to varying WSI sizes, which enhances sequence-independent features while preserving local sequence information, thereby improving sequence information utilization. Extensive experiments demonstrate that Mamba2MIL surpasses state-of-the-art MIL methods. We conducted extensive experiments across multiple datasets, achieving improvements in nearly all performance metrics. Specifically, on the NSCLC dataset, Mamba2MIL achieves a binary tumor classification AUC of 0.9533 and an accuracy of 0.8794. On the BRACS dataset, it achieves a multiclass classification AUC of 0.7986 and an accuracy of 0.4981. The code is available at https://github.com/YuqiZhang-Buaa/Mamba2MIL.

Early dropout prediction requires models that integrate heterogeneous educational records without information leakage. We propose multi-view neural risk augmentation (MVNRA), a cross-fitted framework that separates early-warning predictors into academic, contextual, and digital-engagement views. View-specific encoders and a gated fusion module produce four supervised risk scores. We evaluated MVNRA on a public longitudinal dataset from a Spanish technological university containing 464,739 student–course records and 81 predictors available by December. Within each outer student-level GroupKFold split, the neural risk generator was trained through inner student-grouped cross-fitting. Its out-of-fold fused, academic, contextual, and digital risk scores were then appended to the original predictors before six conventional classifiers were fitted. MVNRA increased mean area under the receiver operating characteristic curve (AUC) for all six classifiers. The largest gains occurred for AdaBoost (0.7717 to 0.9240), linear discriminant analysis (0.8701 to 0.9176), Gaussian Naive Bayes (0.7313 to 0.7786), and Logistic Regression (0.9139 to 0.9352). MVNRA + Random Forest achieved the highest absolute discrimination (AUC $=0.9577$), with increases of 0.0191 in precision–recall AUC and 0.0112 in F1 score. SHAP analysis ranked the fused and academic neural risk scores above the cumulative academic-progress indicators. These findings show that cross-fitted, view-specific neural risk scores can transfer nonlinear structure to conventional tabular classifiers while preserving a SHAP-compatible decision layer and strict student-level separation.