DISCOVER, READ & SHARE
The Institutional Repository is a digital hub for scholarly communication and knowledge dissemination which serves as a vital resource for researchers, students and the wider academic community. Content comprises of research publications, theses, conference proceedings and much more.
Recent Submissions
Multimodal Machine Learning for Maize Disease Detection: A Systematic Review of Architectures and Deployment Challenges
(Mount Kenya University, 2026-02-24) Tonui,Mercy Chepkoech; Kamau,John Wachira; Ongus,Raymond Wafula
Maize diseases continue to threaten agricultural productivity and food
security, particularly in developing regions where early diagnosis remains
constrained by limited expert access. While deep learning has enabled
automated disease detection systems, most existing approaches rely on
unimodal image datasets and cloud-dependent architectures, limiting
robustness and deployment feasibility in resource-constrained environments.
This study presents a structured systematic review of 38 peer-reviewed studies
published between 2020 and 2025, focusing on multimodal machine learning
approaches integrating visual and environmental data for maize disease
detection. Quantitative synthesis reveals that 58% of reviewed studies employ
image-only deep learning models, 26% implement multimodal frameworks,
and only 29% conduct validation under real or semi-real field conditions.
Furthermore, 32% explicitly address deployment considerations, including
edge computing and mobile inference. The findings demonstrate that
multimodal architectures improve robustness and contextual modeling
compared to unimodal systems by integrating phenotypic and environmental
drivers of disease expression. However, increased computational complexity,
synchronization challenges, and limited edge optimization remain barriers to
scalable deployment. This review advances scientific knowledge by providing
a computing-centered synthesis of multimodal architectures, fusion strategies,
deployment constraints, and explainability gaps, identifying key research
priorities in edge efficiency, real-world validation, and interpretable intelligent
systems.
Accuracy of Machine Learning Models for Anomaly Detection in Online Proctored Examinations: A Review of the Potential of Haar Cascades for Mobile Based Proctoring
(International Transactions on Electrical Engineering and Computer Science, 2026-03) Oganda Bartholomew Mogoi; John Kamau; Raymond Ongus
The rapid shift toward digital learning in higher education has made online examinations an everyday reality, increasing the need for reliable and trustworthy anomaly detection systems powered by machine learning (ML). While recent advances in deep learning have improved the technical capabilities of automated proctoring tools, many existing solutions remain difficult to deploy fairly and effectively; particularly in mobile-first environments that are common in low resource settings. Challenges related to computational demands, accessibility, and algorithmic bias continue to limit their practical impact. This review explores the accuracy and suitability of ML models used for anomaly detection in online proctoring, with particular attention given to Haar cascade classifiers; a classical yet computationally efficient approach that remains widely supported on mobile platforms. A structured literature search across seven major databases identified 150 relevant studies, from which 20 met the defined inclusion criteria. The findings show that deep learning models, especially convolutional neural networks (CNNs), consistently achieve the highest detection accuracy but often require substantial computational resources. In contrast, Haar cascades offer fast, low latency detection suitable for mobile devices, although their performance declines under challenging conditions such as poor lighting, pose variation, and facial occlusion. Notably, hybrid approaches that combine Haar cascades with lightweight CNNs emerge as a promising middle ground, balancing efficiency with improved robustness. However, the review also highlights important research gaps, including the scarcity of mobile-centric datasets, limited real world field evaluations, and insufficient testing for fairness across diverse demographic groups. Addressing these gaps will require future research to prioritize mobile optimized model design, standardized evaluation benchmarks, privacy aware computation strategies, and broader empirical validation in authentic exam settings.
Detection of Pregnancy Associated Malaria Among Pregnant Women as a Strategy to Improve Mother Child Health Outcomes in Bungoma County, Kenya
(Mount Kenya University, 2023-11-08) NKONGOLO ,JOSEPH MUKALA; Dr. Dominic Mogere, PHD
Malaria is caused by a protozoa of genus Plasmodium and remains a major public health burden in the Sub-Saharan Africa. In Kenya the prevalence varies between 6.1 to 37% with harmful consequences to both the mother and her baby, including adverse pregnancy outcomes such low birth weight, high morbidity and mortality. However, effective antenatal strategies for improving maternal and child health outcomes through the prevention, early detection and treatment of malaria are still a challenge in resourceconstrained settings. The objective of this study was to detect malaria and to determine its influence on the maternal and the child health outcomes. The response rate was 97%. Malaria test was conducted either via microscopy or rapid test before enrollment, then the cohort splits into malaria positive and negative. The sample size calculation was based on the prevalence of malaria in the unexposed group versus the prevalence of malaria in the exposed group according to the previous studies. Simple random sampling technique was used to enroll participants aged between 18-49 years and having about 16 weeks of gestation. The follow up period ranged from the first antenatal visit (March 2022) and delivery (December 2022). Permission was sought from relevant institutions and informed consent from the participants. Prerequisites on training, pre-testing of tools and standard operating procedures were satisfied. Categorical and outcomes data were analyzed using SPSS 27 and R plotting. Qualitative data were performed via Nvivo computer programs categorized under major themes and sub-themes. Chi-square, Fischer’s exact and relative risk were used for bi-variate analysis at a p-value less or equal 0.05 (95%). The relative risk was 0.999, confidence level of 0.926-1.077. The prevalence of low birth weight was 4.6% with 6 cases of which 3 (4.5%) in the negative cohort and 3 (4.7%) in the positive cohort. Anaemic pregnant women were 41 (31.5%), HIV were 5 (3,8%), pre-eclampsia were 5 (3.8%), gestational diabetes were 2 (1.5%). Otherwise, majority of the participants were aged 18–25 years, were primigravida, were married, had secondary school level, earned between 20-30 thousand shillings, were resident in rural areas, and were in their second trimester. Marital status, gestational age and area of residence were associated with malaria but were not risk factors with a pvalue less than 0.001, 0.001 and 0.028. A panel of sixteen proteins in malaria positive cohort, and six others in malaria negative cohort was identified after computing metadata sample using analysis of variance, t-test and adjusted Bonferroni with a relative influence of biomarkers varying from a mean difference of 2.856690795 to 0.217887462 in malaria positive cohort against -1.185322211 to 0.1622524175 in malaria negative cohort. There was difference in both cohorts with regard to knowledge of side effects p-value <0.01, different doses p-value <0.012 and prior information p-value < 0.003. The results revealed that birth cohort with malaria did not result in significant low birth weight. Therefore, this study recommends to conduct further research for a cost-effective test from the discovered novel biomarkers, which can be useful for low resource settings as an alternative option.
E-Assessment Proctoring Using Artificial Intelligence Technologies: A Review of Practices and Challenges in the African Context
(Mount Kenya University, 2026) Mogoi,Oganda Bartholomew; Kamau,John; Ongus,Raymond
The rapid expansion of e-learning across African higher education institutions has accelerated the adoption of electronic
assessments (e-assessments), intensifying concerns regarding examination integrity. Artificial intelligence (AI)-based proctoring
technologies have emerged as a promising approach to mitigating academic dishonesty through automated monitoring, biometric
authentication, and behavioral analytics. However, the effectiveness, ethical implications, and contextual suitability of these
technologies within the African educational landscape remain underexplored. This review synthesizes empirical and conceptual
studies on AI-enabled e-assessment proctoring in Africa to examine prevailing practices, challenges, and research gaps. Guided by
the PRISMA 2020 guidelines, a systematic search of major academic databases identified 250 relevant studies published between
2015 and 2024, of which 25 met the inclusion criteria for qualitative and quantitative synthesis. The findings reveal a growing
adoption of AI techniques, including facial recognition, keystroke dynamics, gaze tracking, and anomaly detection, alongside
persistent challenges related to internet instability, algorithmic bias, data privacy concerns, system scalability, and institutional
readiness. Notably, there is limited empirical evaluation of mobile-first, low-resource AI proctoring frameworks tailored to African
contexts. Future research should prioritize the development of lightweight, privacy-preserving AI models, incorporate participatory
and inclusive design approaches, and align technological implementations with region-specific regulatory and policy frameworks
to support sustainable and ethical e-assessment practices.
Analysis of business growth strategies on performance of small medium enterprises in Kiambu County, Kenya
(Mount Kenya University, 2025-07) Beatrice Wanjiku Kinyua; Dr. Evans Nyamboga
Performance of small and medium-sized businesses (SMEs) in Kiambu County, Kenya, was investigated in this study in relation to business growth methods. SMEs are well known for their role in generating jobs and advancing the economy as a whole. The study focused on four key growth strategies commonly used by SMEs: market penetration, market development, product development, and diversification. The objective was to determine how each of these strategies influenced SME performance, assessed through indicators such as profitability, market share, customer satisfaction, and business growth. A descriptive research design was adopted for the study. Data were collected from a sample of 370 SMEs through structured questionnaires administered to business owners and senior managers responsible for key functional areas including marketing, operations, finance, and strategic planning. Stratified random sampling was used to ensure fair representation across different sectors. In addition to examining the internal strategies, the study explored the role of external environmental factors such as county government policies, access to financing, and infrastructure in shaping the effectiveness of business growth strategies. The data collected was analyzed using both descriptive and inferential statistical methods. The reliability of the research instrument was confirmed with a Cronbach’s Alpha value of 0.8. Regression analysis revealed that the four business growth strategies collectively explained 61 percent of the variation in SME performance (R² = 0.61). Market development, with a coefficient of 0.071 and a t-value of 2.087, showed the strongest positive influence on SME performance. Product development had a coefficient of 0.062 and a t-value of 2.152, while market penetration had a coefficient of 0.044 and a t-value of 2.925. Diversification demonstrated a modest but statistically significant influence with a coefficient of 0.023 and a t-value of 2.000. Descriptive findings indicated that a majority of SMEs in Kiambu County actively apply business growth strategies to improve performance and competitiveness. Many respondents highlighted the importance of innovation, customer loyalty programs, strategic pricing, and exploring new market segments as essential elements for success. Firms that consistently launched new or enhanced products and engaged in targeted marketing campaigns reported stronger customer retention and increased brand visibility. SMEs that expanded into related markets or adjusted their offerings based on local demand were better positioned to cope with competition and changing customer needs. The study was conducted in compliance with ethical research standards, with approvals obtained from the Ethics Review Committee of Mount Kenya University and the National Commission for Science, Technology and Innovation (NACOSTI). The findings offer valuable insights for SME owners, policymakers, and support institutions seeking to strengthen the sustainability and performance of SMEs. The study provides evidence-based recommendations to support strategic decision-making, guide policy development, and promote long-term SME growth in Kiambu County and other similar regions.
Discover
Top 5 Subjects
- Education 643
- Management 109
- Organization 108
- school of education 104
- Educational planning 93
Top 5 Authors