Research Articles [CoCSE]

Permanent URI for this collection${dspace.ui.url}/handle/20.500.12479/35

Browse

Now showing 1 - 20 of 487

Spatiotemporal patterns of ambient NO2 and NO pollution in Kigali Rwanda
(Springer and Nature, 2026-08-05) Karekezi, Pacifique; Kyeremateng, Kate; Lange, Carissa; Umutoni, Chantal; Kubwimana, Jean; Nimo, James; Wang, Jiayuan; Mottey, Barbara; Gahungu, Paterne; Mirau, Silas; Nyinawumuntu, Claudette; Niyizirugero, Samson; Hakizimana, Pie-Celestin; Mbalawata, Isambi; Ezzati, Majid; Hughes, Allison; Arku, Raphael
Kigali, like many cities in sub-Saharan Africa, faces rapid urban growth alongside the need to manage air pollution and protect public health. Despite its policy efforts, systematic data on nitrogen oxides (NOx: NO2 and NO), key indicators of combustion-related pollution, have been limited. We applied a standardized protocol to characterize city-scale spatial and temporal patterns of NOx across Kigali. Between November 2022 and December 2023, we collected weekly integrated NO2 and NO samples (n = 630 each) at 130 sites representing diverse land-use types. NO2 concentrations ranged from 1.3 to 61.9 µg/m3 (mean 13.9 µg/m3), with annual-equivalent frequently exceeding the WHO annual guideline (10 µg/m3) in urban areas. Exceedances occurred in 39% of sparsely residential, 89% of commercial/industrial, and 99% of densely populated residential sites. NO2 concentrations were significantly higher in urban versus rural areas (18.2 vs. 6.3 µg/m3), near major roads (19.9 vs. 11.6 µg/m3), and at lower elevations (15.4 vs. 9.0 µg/m3). The highest levels were observed in the densely populated districts of Kicukiro and Nyarugenge. Overall, NO2 and NO exhibited strong spatial gradients related to land use, traffic, population density, and topography, highlighting the importance of targeted urban planning and air quality management in rapidly growing cities.
Fractional-order modeling of cholera transmission dynamics with human compliance and environmental reservoirs
(Elsevier Ltd., 2026-07-01) Nyerere, Nkuba; Ruoja, Chiganga; Edward, Stephen
Cholera remains a significant global health challenge, with its transmission dynamics profoundly influenced by human behavior and environmental factors. This study introduces a novel Caputo fractional-order model to capture the memory-dependent effects of human compliance with public health interventions on cholera transmission. By incorporating a behavior-dependent reduction in infection probability, the model accounts for the interaction between public health education coverage and time-varying human behavior, alongside environmental bacterial concentration. The fractional framework improves predictive accuracy by integrating memory effects, reflecting the cumulative impact of past behaviors on the current dynamics of the disease. We establish the qualitative properties of the model, including the existence, uniqueness, and Ulam-Hyers stability of solutions, ensuring robust and biologically meaningful outcomes. The basic reproduction number, derived via the next-generation matrix approach, underscores the critical role of behavioral compliance in disease persistence. The proposed model was validated by comparing its outputs with the observed data, and the goodness of fit was assessed using the coefficient of determination R2, with numerical solutions computed using the Adams–Bashforth-Moulton scheme for enhanced stability. Our findings highlight the synergistic effects of sustained public health education and behavioral adherence in reducing cholera transmission, offering valuable insights for designing effective control strategies. This work advances the mathematical modeling of infectious diseases by integrating fractional calculus and human behavioral dynamics, providing a robust framework for epidemiological analysis and policy formulation.
Enhancing curriculum benchmarking by leveraging NLP: : a case study of higher education in Tanzania
(ACM Digital Library, 2026-07-28) Lukwaro, Elia; Balalusesa, Rogers; Kalegele, Khamisi; Nyambo, Devotha
The fast expansion of higher education institutions and the discrepancy between skill sets and labour market demands have heightened stakeholders' concerns regarding the quality of education. Syllabi are the focal points of this study, as they act as the bridge between education and the skills or competencies required to demonstrate the relationship between the acquired education and the required market competencies. Benchmarking, which has its roots in business, is now widely used in education as a mechanism for evaluating educational metrics and practices and comparing them among institutions or with those of competitors with the aim of improving performance. This paper benchmarks the quality of syllabi using an NLP-based model, namely the sentence bidirectional encoder representations from transformers (SBERT). By utilising the course book to refine the SBERT model, which is a variation of 'all-MiniLM-L6-v2', an experiment is carried out to investigate the most effective parameter metric for model training. With an accuracy score of 92.64%, the model performance score demonstrates a high level of ability to discern conceptual and semantic relationships between sentences, leading to successful syllabi benchmarking outcomes. This work offers a perceptive mechanism to address the mismatch between educationally acquired skills and industry demands.
When Does AI Capability Convert to Advantage? A Complementarity-Gated Framework for Startups in Resource-Constrained Economies
(Journal of Information Technology, Cybersecurity, and Artificial Intelligence, 2026-07-31) Wanyancha, Mwita; Mollel, Moureen
Purpose: This paper develops a conceptual framework linking artificial intelligence (AI) capability, competitive advantage, and startup performance in emerging economies, addressing the limited and fragmented understanding of how AI capability translates into venture outcomes in resource-constrained contexts. Design/methodology/approach: The paper employs a structured review of empirical and theoretical literature, grounded in the resource-based view (RBV) and the dynamic capabilities perspective. It synthesizes evidence across studies to identify points of convergence, divergence, and unresolved tension, and on this basis derives a set of testable propositions. Findings: The review shows that AI capability is a multidimensional construct whose translation into competitive advantage is conditional rather than automatic: it depends on complementary resources such as digital skills, infrastructure, organizational and absorptive capacity, and strategic alignment that are systematically scarce in emerging-economy startups. The paper advances a framework in which competitive advantage mediates the AI capability–performance relationship, but in which that mediating pathway is itself gated by the availability of complementary resources, making the capability–advantage link contingent on context rather than a general regularity. Originality: The paper's contribution is not that AI capability requires complements, a point established in prior literature, but that it formalises where complementarity binds: not on the path from advantage to performance, where most mediation literature locates contingency, but on the prior path from capability to advantage itself. It specifies this as three concrete, measurable gating conditions: digital infrastructure, founder AI literacy, and ecosystem support, rather than treating resource scarcity as unmodelled context. This repositions AI capability from a resource assumed to convert into advantage into one whose conversion is itself the object of theoretical and empirical explanation, with the moderated pathway as the paper's central, testable claim.
When Everyone Has the Same AI: Rethinking Startup Competitive Advantage in the Age of Generative AI
(Journal of Information Technology, Cybersecurity, and Artificial Intelligence, 2026-07-31) Wanyancha, Mwita
Generative artificial intelligence (AI), delivered through general-purpose foundation models accessible on demand at low cost, has become available to firms of all sizes. This accessibility challenges the dominant view in which AI capability, built from scarce data, infrastructure, and talent, is a source of competitive advantage: when the same frontier capability can be rented by any firm, the model satisfies neither the rarity nor the inimitability conditions that the resource-based view requires of an advantage-conferring resource. This paper develops a conceptual framework explaining where competitive advantage resides once generative AI becomes, in effect, a shared utility. Engaging the precedent of the information-technology commoditization debate, it argues that generative AI differs in kind: it commoditizes capability rather than infrastructure, displacing advantage from the technology to the complements it cannot replicate. The paper names this regularity the advantage-relocation mechanism, the displacement of advantage, when a capability-like technology is commoditized, toward co-specialized complements such as proprietary data, domain judgment, customer relationships, and the orchestration capability that converts a common model into a distinctive value proposition. Stated at the level of the firm, with resource-constrained startups as its sharpest boundary condition, the framework yields falsifiable propositions on the relocation of advantage, the role of absorptive capacity in differentiating firms that use identical models, and the accelerated erosion of AI-derived advantage. It contributes a strategic account of competitive advantage under technological commoditization and an agenda for empirical testing.
Machine Learning Techniques to Predict Fetal Nutritional Status
(The Korean Society of Medical Informatics, 2026-07-31) Mduma, Neema; Laizer, Hudson
Objectives Malnutrition remains the leading cause of child mortality in Tanzania, with over 34% of children under 5 years of age affected by stunting and approximately 5% experiencing acute malnutrition. This study aimed to develop a machine learning model to predict fetal nutritional status using maternal and clinical data, thereby enabling early risk identification for health workers and parents and facilitating timely intervention. To enhance practical applicability, the model was deployed within a mobile application to provide accessible, real-time predictions that support prompt clinical and behavioral responses. Methods Using a dataset collected in Tanzania, the performance of multiple binary classification algorithms—logistic regression, multi-layer perceptron, random forest, extreme gradient boosting, and light gradient boosting machine (LightGBM)—was compared using the geometric mean and F-measure. These models were trained on clinical data from 11,703 pregnant women to predict fetal nutritional status based on maternal and clinical variables. Results The results indicated that the LightGBM algorithm achieved the best overall performance in predicting fetal nutritional status. The most influential predictors included maternal age, weight, fetal age, hemoglobin level, number of meals per day, medical history, and education level. Additionally, 93% of respondents reported satisfaction with the application’s predictive functionality, supporting its potential utility for early intervention in low-resource settings. Conclusions These findings highlight the potential of data-driven approaches to address public health challenges in maternal and child health. The proposed model may enable healthcare providers to make timely, informed decisions that improve maternal and fetal outcomes, ultimately contributing to the reduction of child malnutrition in Tanzania.
RGB Image Dataset of White Maize Kernels with Visible Quality Defects for Computer Vision-Based Assessment of Mycotoxin Contamination Risk and Grain Quality
(MDPI, 2026-07-13) Kiwoli, Liston; Nyambo, Devotha; Mgawe, Bonny; Kassim, Neema; Ally, Mussa
Maize (Zea mays L.) is a major staple crop vulnerable to post-harvest deterioration caused by fungal infection and mycotoxin contamination. Visual defects such as discoloration, breakage, insect damage, and mold growth are commonly associated with reduced grain quality and increased contamination risk. This article presents a publicly available RGB image dataset of white maize kernels deposited in Harvard Dataverse for the development of computer vision models for automated grain quality assessment. The dataset contains 5143 high-resolution RGB images acquired using Samsung Galaxy A12 and Samsung Galaxy A54 smartphone cameras under semi-controlled imaging conditions. Images contain either single or multiple kernels and were annotated at the instance level using the YOLO format, resulting in 13,533 labeled kernel instances. Annotations were assigned by experts experienced in mycotoxin-related grain quality inspection. Labels are based solely on visual surface characteristics and do not represent direct chemical measurements of aflatoxins, fumonisins, or other mycotoxins. The dataset provides a practical resource for developing and evaluating machine learning models for maize kernel defect detection, quality screening, and risk-oriented grain inspection applications.
PADLOCK: A Context-Aware On-Device System for Android Permission-Risk Assessment – A Tanzanian Case Study
(Asosiasi Doktor Sistem Informasi Indonesia, 2026-07-26) Mtunga, Ansgar; Mgawe, Bonny; Leo, Judith
Android applications often request sensitive permissions beyond their functional needs, creating privacy and security risks, especially in financial and digital lending apps. Existing Android permission mechanisms provide limited visibility into how permissions are used at runtime. This study proposes PADLOCK, a context-aware prototype for event-driven permission monitoring and contextual risk assessment. The methodology combined a survey-based user study (n = 600), analysis of 3,816 applications, machine learning development, and Android prototype implementation. A compact deep neural network, ShieldAI, was trained on permission-based features with heuristic risk labels to classify applications as benign or potentially higher-risk. The model was integrated into PADLOCK to support on-device monitoring and user-guided permission decisions. On a held-out test set, ShieldAI achieved 96.73% accuracy, 94.82% precision, and 92.42% recall. Controlled on-device testing showed an inference latency of approximately 2–3 ms per prediction and indicated reduced higher-risk permission-granting behaviour in PADLOCK-assisted evaluations. This study contributes an integrated Android prototype combining contextual permission analysis, machine-learning-based risk assessment, and event-driven monitoring. However, the findings remain limited to the evaluated dataset and controlled conditions, and broader real-world validation is required.
Development of information system for enhancing communication between members of parliament and citizens: A study of dodoma, Tanzania
(Elsevier Ltd., 2026-04-12) Mtoi, Mary; Ally, Mussa; Sam, Anael; Mbelwa, Hope
In Tanzania, members of parliament rely on both traditional channels and social media to gather citizens’ concerns. However, limited access to internet-enabled devices, the unavailability of members of parliament, poor attendance at public meetings, and high internet costs create a communication gap in civic participation. Existing civic technology solutions often depend on internet-based platforms, excluding users who rely on feature phones and highlighting a technical gap in prior research and system design. This study develops a context-aware, inclusive digital communication system that bridges infrastructural and technological barriers to foster civic engagement in low-resource environments. Through a literature review, gaps in existing civic technology implementations were identified, and the system was developed using the Extreme Programming Agile Methodology to produce a hybrid platform combining a web application and a two-way messaging system. The resulting Public Participation Information System allows citizens to raise concerns offline via feature phones or online through a bilingual web application in Swahili and English. The system prioritizes accessibility for low-resource users and incorporates features such as availability scheduling, task delegation, and concern tracking. Feedback indicates that it effectively reaches under-connected populations and promotes more inclusive civic engagement. Overall, the system addresses the digital divide in civic communication and offers a scalable framework that integrates low-technology access with modern information systems to support democratic participation in resource-constrained settings.
BiCervi: Pap smear microscopy image dataset for cervical cancer detection and classification
(Elsevier Inc., 2026-07-15) Mbelwa, Hope; Leo, Judith; Kahesa, Crispin; Munema, Asafu; Mkoba, Elizabeth
Cervical cancer is still a major public-health challenge, especially in low and middle-income countries where there are limited numbers of specialists to review Pap smears. While Pap smear screening is widely used, manual reading is slow and can vary from one expert to another, which makes data-driven decision support increasingly important. However, Pap smear imagery datasets from African screening settings are still limited, slowing down the development and fair evaluation of machine-learning models. This article presents a curated dataset of Pap smear microscopy images collected at The Ocean Road Cancer Institute (ORCI) in Tanzania. Images were obtained using a digital microscope with a 10 × /0.25 objective lens, and stored as RGB JPEG images at a resolution of a 2592 × 1944 pixels. The Bethesda System for reporting cervical cytology was used to label the images and organize them into eight diagnostic categories: Atypical Squamous Cells of Undetermined Significance (ASC-US), Atypical Squamous Cells, cannot exclude High-grade Squamous Intraepithelial Lesion (ASC-H), Atypical Glandular Cells (AGC), High-grade Squamous Intraepithelial Lesion (HSIL), Low-grade Squamous Intraepithelial Lesion (LSIL), Negative for Intraepithelial Lesion or Malignancy (NILM), Squamous Cell Carcinoma, and Adenocarcinoma. Labels were independently reviewed by two experts for consistency, and disagreements were resolved through adjudication to reach a final decision; images that could not be labeled with sufficient confidence were excluded during quality control. The resulting dataset comprises 3,000 images and can be used to develop Artificial Intelligence-based tools for cervical cancer detection.
Comparing Dynamic Capabilities in Digital Banking: Economic Integration and Women in the East African Community
(Open Access, 2026) Bhoke Rotente
This study sought to identify the “comparing dynamic capabilities in digital banking institutions in developed and developing countries” through a Quantitative research approach. The population of the study comprised 101 participants who were the IT managers, innovation officers, compliance staff, and executives. They were selected through purposive sampling and Stratified Random Sampling techniques. Questionnaires and interviews were used to collect data. Data were analysed with the aid of the Statistical Package for Social Sciences (SPSS) version 25. The findings of the current study indicate a significant positive correlation with sustainable competitive advantage. Ambidexterity had the strongest correlation (r = .49), followed by UTAUT2 (r = .44), government regulation (r = .41), and digital transformation (r = .38). The strong inter correlations among UTAUT2, ambidexterity, and digital transformation highlight potential overlap in their influence. This means that digital banking in developed countries is more advanced than in developing countries. It is concluded that digital banking in developing countries, Tanzania, particularly, is not satisfied related to developed countries. It is recommended that there is a need to invest in digital technology, particularly for developing countries, to bridge the gap that exists between developed and developing countries.
Late Fusion-Based Multimodal Machine Learning for Driver Fatigue Detection in Emergency Response Driving Scenarios
(Open Access, 2026) Zulfa Sammadile, Deogratias Shidende , Bonny Mgawe , Sabine Moebs
Globally, road accidents are among the leading causes of death and injury, with driver fatigue being one of the main contributors. Driver fatigue is especially critical in emergency response driving, where factors such as long and overnight shifts are common. This study presents a multimodal driver fatigue detection Machine Learning (ML) model that utilizes a late fusion approach. A range of behavioral and physiological data was used to train base learners with various ML models, and their prediction probabilities were used to train four decision tree-based metaclassifiers (RF, XGB, LGB, and HGB) within a stacking late fusion framework. For a broader exploration, ensemble models of these metaclassifiers were trained using simple average, weighted average, majority voting, dynamic weighted average, and best confidence selection methods. Among the metaclassifiers, XGB and dynamic weighted average achieved the highest accuracies of 86.83% and 87.17% respectively, with XGB also yielding the lowest FNR, highlighting its inherent capability in handling heterogeneous and diverse features. To simulate real-world emergency driving conditions, where constant sensor inputs may be intermittent, missing modalities were simulated in Google Colab during training of the metaclassifiers. Later, the robustness of the metaclassifiers was analyzed across various missing modality scenarios, where up to three sensor inputs could be unavailable. In this analysis, XGB achieved the highest stability with the lowest SD of 0.039 across all missing modality scenarios, demonstrating its ability to maintain reliable driver fatigue detection performance even when some sensorinputs are missing.
Gated recurrent unit model for forecasting greenhouse gas concentrations with uncertainty quantification
(Frontiers in Artificial Intelligence, 2026) Erica Hargety Kimei, Devotha Godfrey Nyambo, Neema Mduma, Shubi Felix Kaijage
Reliable and accurate farm-level forecasting of greenhouse gas concentrations from dairy cattle is important to the formulation of climate change miti-gation strategies and policy planning in livestock systems. This study proposed an uncertainty-aware deep learning model that integrated data from multiple sources, including remote sensing and ground-based sensors, to forecast hourly concentrations of nitrous oxide, methane, and carbon dioxide in a controlled zero-grazing dairy system. The forecasting was implemented by formulating a causal multivariate time series using a 24-h lookback window with direct one-step-ahead prediction. A two-stage model evaluation protocol, model hyper-parameter tuning via rolling cross-validation, and holdout Test Evaluation were implemented during model development. The model was evaluated using the coefficient of determination, root-mean-square error, and mean squared error.To evaluate model reliability, the study employed a dual-output gated recurrent unit to estimate the conditional mean and heteroscedastic variance, and Monte Carlo dropout and Gaussian Negative Log-Likelihood to quantify epistemic and aleatoric uncertainty. Results indicate stable generalisation across temporal folds and strong probabilistic calibration, with empirical 95% coverage ranging from 93.6 to 94.8% on the holdout test set. Feature selection indicates that rainfall, normalised difference vegetation index, humidity, temperature, trend, and season influence the prediction of concentration. The study shows that incorporating exogenous variables improves model performance. The proposed framework demonstrates proof of concept for controlled zero-grazing systems, with poten-tial for broader application following multi-site validation.
Dynamics, stability, and bifurcation analysis of a fractional-order HIV–TB co-infection model with environmental transmission
(Elsevier, 2026) Dickson D. Luambano, Mussa A. Stephano, Maranya M. Mayengo
Tuberculosis (TB) and Human Immunodeficiency Virus (HIV) co-infection remain a significant public health challenge, especially in high-burden regions where environmental contamination sustains Mycobacterium tuberculosis. To incorporate memory effects in disease progression and environmental transmission, we develop a fractional-order compartmental model using the Caputo derivative. The model includes an environmental reservoir for TB bacteria and differentiates between HIV, TB, and co-infected stages. We establish positivity and boundedness, derive the basic reproduction number , and conduct stability analysis. Global sensitivity analysis with Latin Hypercube Sampling and Partial Rank Correlation Coefficients highlights key parameters influencing disease spread. The model exhibits backward bifurcation, suggesting that reducing below unity does not necessarily eliminate the disease. Memory effects (lower fractional order) slow disease progression and decrease infection burdens, while environmental transmission considerably sustains disease persistence. HIV transmission rate (), TB transmission rate (), and environmental transmission rate () are the primary drivers of , whereas antiretroviral therapy (), TB treatment (), and environmental clearance rate () are the most effective suppressors. These findings emphasize the importance of integrated intervention strategies that combine direct transmission control, expanded treatment coverage, and environmental sanitation to effectively manage HIV–TB co-infection.
Fowlpox forecasting multimodal dataset from northern Tanzania
(Zenodo, 2026-04-27) Kabeya, Almasi
This dataset contains weather data, epidemiology data, vaccination data, and seasonality data for fowlpox incidence in northern Tanzania.
Optimizing LoRaWAN throughput in maritime environments through adaptive coding and modulation in Rayleigh fading channels
(Taylor & Fransic Online, 2026-06-01) Lyimo, Martine; Mgawe, Bonny; Leo, Judith; Dida, Mussa; Michael, Kisangiri
This article proposes an Adaptive Coding and Modulation (ACM) scheme to improve the throughput of Long-Range Wide Area Network (LoRaWAN) in a maritime environment where wireless communication faces severe challenges from multipath propagation, interference, and Rayleigh fading. Unlike conventional LoRaWAN systems that use fixed physical layer configurations, dynamic adjustment of spreading factor (SF), coding rate (CR), and bandwidth (BW), according to channel conditions based on real-time Signal-to-Noise Ratio (SNR), is another solution to respond to communication challenges in the maritime environment. Simulations using a Rayleigh fading model with six different configuration levels demonstrate that the proposed ACM can achieve a gain of up to 103 times the throughput improvement compared to fixed SF7 under low SNR conditions while maintaining link reliability. This work establishes the foundation for implementing efficient and reliable LoRaWAN networks in maritime Internet of Things (IoT) applications.
Can dual active ingredient Interceptor® G2 insecticide-treated net (ITN) replace indoor residual spraying (IRS) efficiently? A case study in Sakassou, Côte d’Ivoire
(Springer Nature, 2026-05-28) Chabi, Joseph; Shirima, Gloria; Masanja, Brian; Sylvester; N’Guessan; N’Guessan Gbalegba, Constant; Kouassi, Bernard; N’Guessan Broudje, Brice; Ako Edi, Constant; Yokoly, Firmain; Adimi, William; Kouame, Ruth-Marie; Anian, Valentin; Kyerematen, Rosina; Egyir-Yawson, Alexander; Kiware, Samson; Dadzie, Samuel
Background Following three successful rounds of indoor residual spraying (IRS) implementation in the district of Sakassou (Côte d’Ivoire), IRS was withdrawn and replaced by insecticide-treated nets (ITNs). This study evaluated the entomological and epidemiological impacts of Interceptor (IG2) ITNs distributed in Sakassou to determine whether the protection offered by IG2 ITNs was adequate to suppress malaria transmission post-IRS withdrawal. Methods This study is a quasi-experimental evaluation using historical data and additional data collections on entomological indicators and malaria incidence to assess malaria transmission trends in Sakassou. The vector control optimization model was adapted to evaluate the effectiveness of the IRS and IG2 ITN deployment on malaria transmission dynamics. Additionally, we used an interrupted time series model to analyze routinely reported malaria cases in the District Health Information Management System (DHIS2) to determine the epidemiological impact of both interventions. Counterfactual trends were generated for the post-IRS withdrawal period during which IG2 ITNs were distributed. Results The results showed a 55.4% (95% CI 48.3–62.4%) reduction in the human biting rate (HBR) when IRS was deployed and 48.8% (95% CI 42.8–54.6%) when IG2 ITNs were distributed, compared with standard pyrethroid-only nets. No statistical difference was recorded between the HBR of IG2 ITNs and IRS (P = 0.164) during the implementation of IG2 ITNs and the counterfactual of IRS. Similarly, IRS resulted in a 64.7% (95% CI 56.6–72.8%) decline in EIR while IG2 ITNs resulted in a comparable reduction of 61.9% (95% CI 54.2–69.6% ;P = 0.616) over the same period. Furthermore, a 26% reduction in malaria cases was recorded immediately after spraying (IRR = 1.02; 95% CI 1.00–1.04; P = 0.005) with cumulative impact over time and spray rounds and was similar to IG2 performance (IRR = 1.03; CI 1.00–1.07; P = 0.040). Conclusions The study findings suggest that IG2 ITNs provided entomological efficacy comparable to clothianidin-based IRS but could not adequately suppress malaria cases after IRS withdrawal, possibly due to plastic-vector feeding behavior and late timing of IG2 deployment. Overall, the study shows how the dynamics of malaria transmission and operational decisions could impact the effectiveness of both IRS and ITNs as vector control tools. These findings provide key information for malaria programs and policymakers to consider when deploying vector control interventions as countries work toward malaria elimination.
Chance constrained nonlinear model predictive control for multiple systems with shared resources
(Elsevier B.V., 2026-06-04) Mugenga, Ineza; Geletu, Abebe; Mirau, Silas; Li, Pu
This paper proposes chance constrained nonlinear model predictive control (CCNMPC) for multiple interconnected systems sharing limited resources under uncertainty. Each subsystem with uncertainties has its local objectives and constraints and is coupled to the other subsystems through the shared resource. The shared resource limitation is satisfied by a global chance constraint which is decomposed into individual local chance constraints via the Bonferroni inequality, enabling a distributed solution approach. The chance constrained problem is then transformed into a sequence of smooth deterministic problems through the inner–outer approximation method, in which the smoothing parameter is shown to act as an implicit constraint-tightening mechanism that recovers the classical Gaussian tube-MPC back-off in the zero-smoothing limit. Within this framework, we prove probabilistic one-step recursive feasibility under unbounded disturbances. The resulting distributed nonconvex problem is solved using an adapted Augmented Lagrangian based Alternating Direction Inexact Newton (ALADIN) algorithm. The framework is validated on a shared battery energy storage system serving multiple microgrids with non-Gaussian demand forecast errors over a 24-h horizon: the empirical violation rates remain below the prescribed 5% threshold and the terminal invariance condition is satisfied with substantial margin throughout.
A labelled dataset of healthy and diseased common bean (Phaseolus vulgaris) from Tanzania
(Elsevier Inc., 2026-05-31) Mduma, Neema; Laizer, Hudson
Common bean (Phaseolus vulgaris) is an important food and cash crop in Tanzania, where it contributes to household nutrition and income among smallholder farmers. Its production is, however, constrained by diseases such as bean rust and bean anthracnose, which can cause substantial yield losses and reduce crop quality. This article presents a labelled image dataset of common bean leaves collected under field conditions in Northern Tanzania to support research and application development in computer vision, machine learning and digital crop health. The dataset comprises 155,842 labelled multi-season images belonging to three classes: healthy leaves, bean rust and bean anthracnose. Images were acquired from farms in Arusha, Kilimanjaro and Manyara regions using smartphone cameras and were subsequently reviewed and validated with support from agricultural extension officers and plant pathologists to improve annotation reliability. The dataset is organized into class-specific compressed files and is publicly available through the Zenodo repository. By providing a large field-based image resource captured under variable real-world conditions, this dataset can support the development, training and evaluation of image-based models for common bean disease identification.
Macroeconomic transmission effects on stock market performance in Tanzania: evidence from a structural VAR analysis
(Frontiers, 2026-05-20) Peter, Michael; Mirau, Silas; Sinkwembe, Emmanuel; Kasumo, Christian; Guambe, Calisto
This study examines the dynamic interactions between key macroeconomic indicators and stock market performance in Tanzania, a frontier market in sub-Saharan Africa. Using a Structural Vector Autoregression (SVAR) framework, we analyze monthly data for the Tanzania Share Index (TSI), gold returns, inflation rate, and electricity consumption employed as a high-frequency proxy for real economic activity to address the temporal limitations of quarterly GDP data common in developing economies. Johansen co-integration analysis identifies four stable long-run relationships, while a one-lag VAR specification suggests rapid adjustment dynamics. Impulse response functions reveal that shocks dissipate within three to five months. Notably, the TSI responds negatively to economic activity shocks, indicating a structural mismatch between stock market valuations and real-sector growth. Forecast error variance decomposition further underscores this decoupling: own shocks explain approximately 75%–80% of TSI fluctuations, suggesting that within this four-variable system, macroeconomic factors account for a modest share of equity variation. Gold returns, however, exhibit persistent positive responses to inflation shocks, confirming their role as an inflation hedge. Overall, the findings suggest Tanzania's financial market remains relatively disconnected from real-sector fundamentals, reflecting its early stage of financial deepening. Policy recommendations emphasize strengthening financial infrastructure, enhancing macro-financial linkages, and developing robust commodity-price monitoring systems to improve signal transmission to capital markets.

Browse

Recent Submissions