Accurate white blood cell classification is an important component of hematological image analysis, yet robust automation remains challenging because microscopic smear images often exhibit substantial morphological similarity across subtypes, staining variation, acquisition heterogeneity, and sensitivity to surrounding background content. Existing single-backbone approaches may not fully capture the complementary local texture, cellular morphology, and global contextual information required for reliable recognition across heterogeneous white blood cell images, while the effects of preprocessing choices such as cropping and image standardization are often insufficiently examined. To address these issues, this study proposes a three-branch hybrid fusion architecture for five-class white blood cell classification on a combined PBC-Normal and Raabin-WBC dataset comprising 27,606 images, including 3337 basophils, 6158 eosinophils, 4685 lymphocytes, 4,108 monocytes, and 9318 neutrophils. The framework integrates MobileNetV3, DenseNet121, and pretrained DINOv2 feature branches to capture lightweight convolutional cues, densely reused local representations, and transformer-based global semantic features, respectively. Branch-wise 512-dimensional embeddings are combined through feature stacking, layer normalization, multi-head attention, and a residual-style classification head for final five-class prediction. Performance was examined under four image-preparation settings, namely not cropped and unprocessed, not cropped and processed, cropped and unprocessed, and cropped and processed, together with dataset-specific analyses on PBC-Normal and Raabin-WBC. Across these comparative experiments, the proposed model achieved the strongest overall performance, with particularly high results on the PBC-Normal dataset and consistently favorable behavior across the combined-dataset evaluation settings, while processed inputs generally improved convergence and classification quality relative to unprocessed counterparts. These findings indicate that attention-guided CNN-transformer fusion, combined with controlled preprocessing evaluation, provides an effective and robust framework for automated five-class white blood cell classification across heterogeneous microscopic images.
Zimbabwe is a remittance-dependent society and uncertainty has become a constrain driver to the inflows of remittances due to the overhanging institutional trust. Domestic inflation and natural disasters in Zimbabwe, as well as global shocks such as COVID-19 and the Russia-Ukraine war, have escalated economic instability which adversely impact on remittance flow into Zimbabwe. The study's objective is to establish the effects of uncertainty on remittances in Zimbabwe using data for 2000–2019. The study employed the ARDL bounds and GARCH model. The ARDL results show a long-run relationship between remittances and uncertainty with high speed of adjustment 85.1
In this study, we explore factors affecting the intention of Generation Z tourists to use smartphone apps in tourism marketing, based on the Unified Theory of Acceptance and Use of Technology (UTAUT) model. A voluntary, online-administered survey that was promoted through social media and email provided 530 responses, 399 of which were analyzed. Findings of the study, which were obtained with the help of Smart PLS 3.2.9 and SPSS 27 to conduct structural equation modeling, show that the performance expectancy, effort expectancy, social influence, enjoyment, and facilitating condition exert a significant effect on the behavioral intention of Gen Z tourists. Practical implications. The findings present accessible, organized information that can be utilized by individuals working in the tourism sector to improve their practices when it comes to digital interaction with tourism in terms of Gen Z. Originality/value. The research extends the UTAUT model to the Gen Z and tourism context, examining the adoption of the smartphone application in tourism among Gen Z. Existing UTAUT2 studies assume utilitarian adoption, but Gen Z tourism app use is driven by hedonic and experiential motivations, creating a mismatch between technology adoption theory and experiential tourism behavior.
Mobile money interoperability (MMI) has reshaped Ghana’s digital financial landscape, yet its effects on individual users’ everyday finances remain poorly understood. While aggregate transaction data paint an encouraging picture, total interoperable transaction value grew from GH₵26.4 billion in 2022 to GH₵32.8 billion in 2023. This growth has not been systematically translated into evidence about whether ordinary users are saving more, spending less on transaction fees, accessing credit, or becoming more financially resilient. This study employs a mixed-methods design, combining a structured survey of 350 adult mobile money users from the Greater Accra, Ashanti, and Bono regions, conducted through stratified random sampling, with semi-structured interviews with 20 purposively selected participants. Quantitative data are analysed using binary and ordinal logistic regression in IBM SPSS v26. Qualitative data are thematically coded and triangulated with quantitative findings. MMI usage is associated with reduced cross-network transaction friction and with increased savings frequency, broader access to algorithmic digital credit, and greater household financial resilience. However, these associations are moderated by financial literacy, geographic location, and unresolved provider-level interchange fee disputes, which partially erode user-level cost savings. Interoperability is a meaningful but incomplete driver of financial well-being at the individual level. Realising its full potential requires complementary investment in financial literacy, expansion of agent networks in rural areas, and regulatory reform of interchange fee structures. These findings carry implications for FinTech policymakers across Sub-Saharan Africa.
Aspect-level sentiment analysis of e-commerce reviews is important for identifying fine-grained customer experience signals that are often hidden by overall star ratings. This study proposes a dual-track framework for USA-based Amazon customer reviews, combining text-driven multi-aspect sentiment classification with structured-feature satisfaction prediction. For the text-based task, reviews are modeled across three operational aspects: Product Value, Delivery Fulfillment, and Service Usability, each using a three-class sentiment scheme of Negative, Neutral, and Positive. The proposed deep learning model integrates DeBERTa-v3 contextual embeddings, bidirectional recurrent refinement, and aspect-specific additive attention to separate overlapping sentiment cues within the same review. Three recurrent variants, Bi-GRU, Bi-RNN, and Bi-LSTM, are evaluated under the same modeling framework. On the held-out Amazon Customer Reviews test set, the DeBERTa-v3 + Bi-GRU + aspect-attention model achieves the best overall performance, with 0.737 accuracy, 0.757 macro-F1, and 0.760 weighted-F1, outperforming the Bi-LSTM and Bi-RNN variants. Ablation analysis further shows that sequential refinement and aspect-specific attention improve performance over encoder-only and shared-attention configurations. In the structured-feature satisfaction task, LightGBM achieves the strongest result among the evaluated machine learning models, with 82.37
This paper presents a computational refinement of classical optimality procedures for solving Transportation Problems (TPs). Rather than altering the underlying linear programming formulation, the method introduces a structured loop-enumeration strategy within the transportation simplex framework. Specifically, admissible improvement loops are generated from the current basis (occupied cells) and constrained to include exactly one nonbasic (unoccupied) cell, reducing redundant loop exploration. A formal analysis establishes feasibility preservation, strict descent when an improving loop exists, finite termination, and equivalence with classical optimality conditions via reduced costs. Computational experiments on benchmark problems from literature compare the new method with the Stepping–Stone and MODI methods. To evaluate computational efficiency, all computational experiments were conducted using MATLAB R2022b on a computer equipped with an Intel Core i7 processor (3.20 GHz), 16 GB RAM, running Windows 10 (64-bit), Execution time was measured using MATLAB’s built-in timing functions. Each method was run five times per dataset, and the average CPU execution time (in seconds) was recorded to ensure consistency. Results show that the new approach yields optimal solutions identical to classical methods and achieves comparable or improved performance on average runtime, with clearer benefits on medium-to-large instances where loop redundancy is more pronounced. For small instances, differences are minor and may vary due to implementation overhead. The findings position the method as a practical, verifiable computational refinement for efficient loop selection in transportation simplex implementations.
We propose a fractional-order multi-host epidemic model that captures zoonotic spillover dynamics across human, pig, and bird populations. The model is formulated using the Caputo fractional derivative to incorporate memory effects arising from past infection states, which are not accounted for in classical integer-order models. The transmission structure explicitly represents cross-species pathways, including bird-to-pig, pig-to-human, and within-host transmission dynamics, providing a realistic framework for multi-host disease spread. Although the model is general, it is motivated by zoonotic systems such as Nipah virus, where such transmission routes are epidemiologically relevant. Unlike existing studies, this work simultaneously integrates fractional dynamics, multi-host coupling, and time-dependent optimal control strategies, thereby extending both classical epidemic models and recent fractional-order formulations. Numerical simulations show that reducing the fractional order significantly lowers peak infection levels (by more than 40
This study examines the disruptors of vaccine supply chains (VSCs) that underpin mass vaccination campaigns during a global pandemic. Employing an exploratory qualitative content analysis, the study identifies critical VSC flows and the causal factors that disrupted those flows during the COVID-19 mass vaccination campaign. Data were extracted from white papers, reports, academic papers, and publicly available expert presentations involving more than 100 professionals from sectors relevant to vaccine production, distribution, regulation, logistics, and administration. The findings reveal seven critical flows required for rapid mass vaccination: product, information, funds, vaccine recipients, knowledge, skilled workers, and waste. Across these flows, the analysis identified 116 main and sub-causal disruptors, including shortages of manufacturing inputs, cold-chain and logistics capacity constraints, inadequate real-time data visibility, financing delays, technology-transfer barriers, skilled labor shortages, vaccine hesitancy, and waste-management limitations. The study contributes to the literature by providing a holistic, structured taxonomy of pandemic VSC disruptors grounded in the COVID-19 experience. Practically, the findings provide a high-level diagnostic framework for public health agencies, manufacturers, logistics providers, and policymakers involved in planning and managing mass vaccination supply chains during future pandemics.
The study investigates hotel workers' acceptance of Robotic Process Automation (RPA) in Jaipur, India, using the extended Unified Theory of Acceptance and Use of Technology (UTAUT-3) framework. In recent times, with automation technologies gaining momentum in the hospitality industry for the enhancement of operational effectiveness and service delivery, understanding the factors affecting employee acceptance has assumed critical importance. The researchers adopted a quantitative approach, where data was collected from 170 employees from various hotel categories and departments. The data was analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) to test the hypothesized relationships. The results confirmed all the proposed relationships in the UTAUT-3 model, with Performance Expectancy (β = 0.325) being the strongest predictor of Behavioral Intentions, followed by Effort Expectancy (β = 0.248). The model explained a substantial variance of 68.5
Modern supply chains require decision-support systems that integrate operational efficiency, financial performance, and environmental sustainability. However, existing studies often examine blockchain transparency, machine learning forecasting, and optimization models separately, with limited integration for real-time supply chain decision-making. This study proposes an integrated framework combining machine learning prediction, multi-objective optimization, and blockchain-enabled traceability to improve supply chain performance and sustainability. The methodology includes data preprocessing, predictive modeling using a Random Forest algorithm, and multi-objective optimization based on Linear Programming to simultaneously minimize operational costs and carbon emissions. A blockchain-based transaction layer is incorporated to enhance supply chain transparency and data integrity. Empirical analysis using supply chain operational and sustainability indicators demonstrates strong predictive performance (R2 = 0.893; RMSE = 0.503) and reveals a measurable trade-off between cost efficiency and carbon emissions across supplier–buyer allocations. The proposed framework provides a structured approach for integrating predictive analytics, sustainability-oriented optimization, and blockchain-supported traceability, offering insights for improving transparency, efficiency, and sustainability in modern supply chains.
While the f-chart method is a standard tool for designing solar thermal systems, a significant research gap exists regarding the precise impact of the heat exchanger correction factor (F′R/FR) under different operational conditions. Most studies assume a near-unity factor, potentially underestimating the required collector area when a heat exchanger is present. This study is motivated by the need for a more rigorous approach to sizing systems that incorporate heat exchangers. We propose a methodology that utilizes monthly mean solar irradiance, collector loop flow rates and typical temperatures to estimate the correction factor. Our findings reveal that in worst-case scenarios, the required collector area can increase by 12–37
This paper examines the distinct effects of institutional regulation and broader institutional quality on agricultural innovation across 48 Sub-Saharan African (SSA) countries over 2000–2021, with attention to structural heterogeneity and coevolutionary dynamics. We employ a triangulated empirical strategy combining within-country Fixed Effects (FE), System GMM, and PLS-SEM. Agricultural innovation is measured through a custom six-component WIPO/GII composite index. Governance is decomposed into a regulatory dimension (Rule of Law and Regulatory Quality: Regulation index) and a broader institutional quality dimension (Voice, Political Stability, Government Effectiveness, and Corruption Control: Qualityinstit index). FE finds no significant short-run within-country effect, consistent with near-unit-root innovation persistence (ρ = 0.981) and known institutional adjustment lags. System GMM identifies a positive long-run structural effect (β = 0.029, p > 0.10 aggregate; significant when governance dimensions are tested via regional sub-samples). Regional decomposition uncovers a governance transition pathway: institutional regulation drives innovation exclusively in Southern Africa (β = 0.775, p = 0.041), while broader institutional quality is the binding constraint in Central Africa (β = 0.548, p = 0.011). PLS-SEM confirms bidirectional coevolution (β = 0.604, p < 0.001 in both directions). Universal innovation decline across SSA (2000–2021) is five times larger in less rural economies (− 2.95 SD) than in highly rural ones (− 0.57 SD). Context-sensitive institutional policies must account for sub-regional governance trajectories. Countries at early institutional stages should prioritize political stability and government effectiveness before investing in regulatory reforms; more advanced economies can leverage regulatory quality for direct innovation returns. The bidirectional relationship implies that innovation investments can themselves strengthen institutional quality, opening a virtuous cycle that current policies overlook.
This study investigates the influence of the Big Five personality traits on loan repayment intention among microfinance borrowers in Tanzania. While microfinance institutions (MFIs) play a crucial role in promoting financial inclusion and poverty reduction, loan defaults remain a significant challenge threatening their sustainability and outreach. Traditional borrower assessments often neglect personality factors that can drive repayment intention and behavior. Drawing on the Big Five personality theory and the theory of planned behavior, this research examines how agreeableness, conscientiousness, extraversion, openness to experience, and neuroticism affect repayment intention. A survey of 900 microfinance clients was conducted, and data were analyzed using partial least squares structural equation modelling (PLS-SEM). The results reveal that conscientiousness and openness positively and significantly influence loan repayment intention, whereas extraversion and neuroticism have negative and insignificant effects on loan repayment intention. Agreeableness positively but insignificantly affects loan repayment intention. These findings highlight the contextual complexity of personality’s role in loan repayment and suggest the value of integrating personality assessments into microfinance lending practices. The study contributes to behavioral finance theory and offers practical and policy recommendations for enhancing loan repayment and hence promoting sustainable microfinance in developing economies like Tanzania.
Automated pharyngitis recognition from RGB throat images can support rapid screening and triage, but reliable classification is difficult due to substantial intra-class variability, illumination and viewpoint changes, and strong visual similarity between mild inflammation and healthy appearances. To address these challenges, we propose SeAttnFusionNet, a dual-branch feature extraction network that learns complementary evidence and performs feature-map level fusion before GlobalAveragePooling2D, followed by a compact Dense, Dropout, Dense, Output head. The texture branch refines fine-grained cues using SeparableConv2D, Batch Normalization, and stacked Squeeze-and-Excitation blocks, while the context branch captures structural patterns using Multi-scale convolutions followed by Resize, Concatenate, Normalize, Local Window Attention, a Dynamic Gating Mechanism with Element-wise Multiplication, and Cross Scale Feature Fusion through GAP, Activation, and Sigmoid. The two branch outputs are concatenated and pooled only after fusion, preserving spatially resolved complementary representations prior to global aggregation. We evaluate the model using leakage-safe 5-fold cross-validation with a per-fold 80/20 split on 312 samples consisting of 185 no pharyngitis and 127 pharyngitis cases, with each fold using 250 training samples (148 no pharyngitis, 102 pharyngitis) and a 62-sample test fold (37 no pharyngitis, 25 pharyngitis) at an input size of 224 × 224 pixels. With batch size 32, 50 epochs, Adam optimizer, learning rate 0.0001 scheduled by ReduceLROnPlateau, categorical cross-entropy loss, and regularization via batch normalization and dropout 0.35, SeAttnFusionNet achieves 97.69
This paper develops a dual-risk valuation theory that extends the classical discounted cash flow model by adding an analytically distinct source of risk: expectational risk. We model expectational risk through the valuation ratio K , which measures systematic distortion in projected cash flows (expectational distortion), and study its interaction with the benchmark discount rate c (market-based discounting). The objective is to embed ex-ante alpha into valuation as a structural quantity rather than treat it as an ex-post residual. Methodologically, we build an analytical framework that defines the gain of capital g = f(c, K) and the effective cost of capital c_eff as the internal rate that reconciles distorted expectations with fair value. We derive closed-form conditions under which discounting ceases to be equivalent to expected return, characterize the horizon behavior of c_eff , and establish asymptotic results for the per-period equivalent of g , denoted γ . The main findings show that K and c jointly determine valuation outcomes, with g providing a forward-looking measure of ex-ante alpha whenever systematic distortion persists. We prove conditions where c_eff c , identify regimes in which short and medium horizons are most sensitive to K , and show that γ→ 0 as the horizon lengthens, which clarifies when the expectational distortion captured by K becomes irrelevant for intrinsic value. These results reframe alpha as a structural feature of biased expectations and separate the two risk primitives: market opportunity cost (market risk) and expectational distortion (expectational risk). The framework generalizes DCF to incorporate systematic errors in cash-flow projections and yields practical guidance for enterprise valuation.
This article introduces the venture-ready innovation quotient (VRIQ), a structured multidimensional framework designed to bridge the persistent gap between academic theory and venture capital (VC) due diligence practice. Grounded in foundational management and economic theories—the resource-based view, dynamic capabilities theory, signaling theory, and agency theory—the VRIQ operationalizes innovation assessment for early-stage ventures through a behaviorally anchored scoring rubric. Its principal contribution is a departure from linear checklists and single-composite-score models toward a diagnostic profile that characterizes a venture’s strengths, weaknesses, and risk configuration across three core dimensions: potential, execution, and impact. Preliminary psychometric evidence from a pilot study ( n = 35 ) yields an inter-rater reliability of ICC(2,k) = 0.85 and Cronbach’s α≥ 0.79 across all dimensions. For practitioners, the VRIQ provides a structured due diligence and risk management instrument; for scholars, it advances a theoretically grounded and empirically testable research agenda.
Abstract Selecting an appropriate supportive response for a mental health question post is a key step toward scalable mental health question answering, yet it remains difficult due to semantic mismatch, noisy informal language, and the need to align supportive intent beyond surface lexical overlap. We propose MF-GAT, a novel multi view graph attention matching framework that constructs a Concept Interaction Graph (CIG) to explicitly encode post response concept alignments and their interaction structure. MF-GAT learns three complementary evidence streams, including local interaction features, multi view fusion features, and global context features, and applies view specific graph attention to propagate and reweight informative relational signals over the CIG. A gated fusion module then adaptively integrates the view representations into a unified matching vector for prediction. We evaluate MF-GAT both as a pair classification model and as a retrieval ranking model for selecting the best support from a candidate pool, reporting Accuracy and F1 together with standard ranking metrics including MRR and NDCG. On the MHQA benchmark, MF-GAT achieves 0.95 Accuracy and 0.85 F1, outperforming BERT (0.89, 0.67), CIG-GCN (0.92, 0.79), ARC-II (0.85, 0.60), and MatchPyramid (0.82, 0.59). These results show that novel multi view interaction graph modeling with attention based propagation improves both supportive response classification and practical retrieval quality for mental health support selection.
Abstract Accurate risk classification in automobile insurance claim frequency in emerging markets remains a critical challenge for actuarial practice and regulatory oversight. Claim data are characterized by sparsity, overdispersion, and structural zeros, reflecting low-frequency claim occurrences and heterogeneous risk exposures across regions. The study compares count models for sparse automobile insurance claims and developed a Bayesian hierarchical Negative Binomial (NB) Hurdle model using 10,149 policies from Tanzania (2014–2024). Poisson models were found inadequate due to their restrictive equidispersion assumption, while the NB regression effectively addressed overdispersion. To further account for excess zeros and regional heterogeneity, two-part Hurdle models, including Bayesian NB Hurdle, were employed. Model comparisons using log-likelihood, AIC, BIC, and cross-validation demonstrated that the Bayesian NB Hurdle model provided superior fit and predictive accuracy. Results reveal significant heterogeneity in claim behaviour across Tanzania. Dar es Salaam showed the highest claim frequencies, followed by Arusha, while Mbeya, Dodoma, and Mwanza exhibited lower claims, highlighting clear spatial variation in insurance risk. Covariates such as policy type, driver age, and vehicle age showed minimal influence on claim frequency. These results underscore the importance of accounting for regional heterogeneity in sparse claim data and demonstrates that the Bayesian NB Hurdle model effectively addresses both overdispersion and zero inflation, delivering robust and regionally sensitive predictions for automobile insurance claims. The findings further indicate that sparse stochastic systems in emerging insurance markets are best modelled using flexible two-part count frameworks, which provide reliable probabilistic forecasts, enhance risk segmentation, and support more informed actuarial decision-making.
This study reveals the impact of digital maturities on customer relationship management (CRM) performance in automotive industries. A fuzzy comprehensive evaluation (FCE) approach is employed to evaluate CRM performance across two enterprise cohorts exhibiting varying levels of digital maturities. Empirical results demonstrate that the application of digital technologies has a significant impact on CRM, but there are not significant disparities in the final FCE grades between two groups, underlying the strategic importance of digital technologies for business to maintain sustainable innovation. However, the first-class index “Increase Customer Satisfaction” is rated as rank “Excellent” in AAA-level certified enterprises while “Good” rank is rated for AA-level below certified enterprises, this indicates the existence of misalignment between advanced information systems tools and immature business processes in enterprises with lower digital maturities. Practical implications for policy makers include the promotion of digital maturity assessment before introducing universal digital platforms. For practitioners, it is imperative to have a business process maturity assessment prior to significant digital platform investments.
Abstract Contractor-related deficiencies remain a critical challenge in workplace health and safety management within high-risk industries, particularly in the oil and gas sector, where heterogeneous safety practices and fragmented oversight mechanisms undermine effective risk control. Existing contractor evaluation approaches often rely on checklist-based or lagging indicators, offering limited ability to capture interdependencies among safety dimensions or to differentiate contractor performance in a meaningful and decision-relevant manner. This study develops and validates the Contractor Health, Safety, Environment, and Quality Performance Index (CHPI) as an integrated, evidence-based framework for systematic contractor performance evaluation. The study adopted a multi-method approach integrating expert judgment and archival compliance data. Health and safety indicators were identified through literature review and expert consultation, refined using Content Validity Index assessment, and weighted using a Fuzzy Analytic Network Process to capture interdependencies. The resulting weights were combined with standardized contractor records to compute CHPI scores. Robustness was confirmed through sensitivity analysis demonstrating stable contractor rankings, while a Random Forest-based analysis was used as a complementary validation to assess alignment between expert-based weights and data-driven importance. The results show that the CHPI enables nuanced differentiation of contractor performance, supports pre-contract screening and targeted intervention strategies, and enhances transparency in performance-based regulation. By integrating interdependency-aware weighting with empirical validation, the CHPI provides a scalable and adaptive decision-support tool that can strengthen contractor governance, improve safety performance monitoring, and support societal progress through more accountable risk management in high-risk industrial environments.
