Welcome to NM-AIST Research repository

This is a digital service that collects, preserves, and distributes digital material. Repositories are important tools for preserving an organization's legacy; they facilitate digital preservation and scholarly communication.

Our Communities

Select a community to browse its collections.

Now showing 1 - 5 of 6

Recent Submissions

Item

Development of a mobile application for detecting cervical cancer using deep learning model

(NM-AIST, 2025-07) Kiswaga, Dorcas

Advancements in technology are significantly enhancing healthcare, particularly in regions with limited clinical expertise, such as Tanzania. Cervical cancer remains a major public health challenge in these areas, where access to regular screening is often inadequate. This project aimed to develop a mobile application powered by deep learning to assist pathologists in detecting cervical cancer without relying on a microscope or physical observation. Traditional diagnostic methods like Pap smears, HPV testing, and colposcopy can be error-prone, especially in resource-limited settings. The dataset comprised 2,276 histological images of cervical tissue, categorized into Grade One, Grade Two, Grade Three, and Not Cancer. These images were collected with permission from SoftMed (T) Limited, a private pathology laboratory in Arusha, Tanzania, and were annotated by qualified pathologists to ensure clinical accuracy. The Design Science Research Methodology (DSRM) was adopted to guide the development and evaluation of the solution. A convolutional neural network (CNN) was implemented for image classification, and several pre-trained models ResNet50, VGG19, MobileNet, and a custom CNN were evaluated. MobileNet was selected due to its high performance and lightweight architecture, ideal for mobile deployment. The final model achieved 99.59% training accuracy and 89.6% validation accuracy. The trained model was integrated into an Android mobile application, enabling healthcare providers to upload and analyze cervical histology images directly from their devices. This application offers a fast, reliable, and accessible diagnostic tool, enhancing cervical cancer screening in underserved areas. The project demonstrates how deep learning and mobile technology can bridge critical gaps in healthcare delivery.

Item

Exploratory conversations with biodiversity‑oriented civil society groups on the potential applications of gene drive‑modified mosquitoes for malaria control in Tanzania

(Springer Nature, 2026-01-13) Finda, Marceline; Sambo, Maganga; Malika, Goodluck; Njalambaha, Rukiya; Salum, Sheikha; Okumu, Fredros

Gene drive-modified mosquitoes (GDMMs) are gaining attention as sustainable tools to complement existing malaria control strategies. Their ability to self-propagate and spread through wild mosquito populations offers the promise of low-cost, long-lasting impact, but also raises ecological, ethical, and governance concerns. In this evolving debate, civil society organizations (CSOs) are pivotal actors in shaping dialogue, representing community concerns, and influencing policy decisions. This study examined the perspectives and recommendations of biodiversity-oriented CSOs on the governance, testing, and potential application of GDMMs for malaria control in Tanzania. An exploratory qualitative design was employed, involving eight in-depth interviews, one focus group discussion, and three large group discussions with representatives from ten biodiversity-focused CSOs in Tanzania. Participants were selected purposively based on prior involvement in national or regional dialogues related to biotechnology; and the discussions focused on concerns, uncertainties and needs associated with testing and potential use of GDMMs for malaria control, as well as the balance of prospective benefits against long-term environmental risks. Transcripts were analyzed thematically using NVivo 12 Plus. Participants expressed cautious support for research on GDMMs for malaria control but raised concerns about scientific uncertainty, limited local expertise, inadequate transparency, potential transboundary effects and technological dependency. They emphasized the importance of generating robust, context-specific evidence before considering any environmental releases of gene drives; and highlighted concerns over inadequate accountability, particularly the lack of clarity on who would assume responsibility if adverse outcomes arise. They also advocated for early, inclusive, transparent, and continuous engagement with both target communities and the broader public. Lastly, to ensure objective and impartial oversight, they recommended development of local expertise that is independent of technology developers and sponsors. The CSOs’ perspectives were diverse but broadly aligned with the precautionary principle, calling for preventive action amid uncertainty, clear accountability, and the pursuit of safer alternatives. Although many expressed serious reservations about gene drive mosquitoes, there was a shared recognition that research on the technology is necessary, provided it is conducted under controlled, transparent, and auditable conditions. Overall, these exploratory discussions underscored the need for: (i) balanced dialogue between advocates and skeptics, (ii) robust ethical and regulatory frameworks covering the full life cycle of the technology, (iii) sustained community and stakeholder engagement from the early stages of research and development, (iv) enhancements of in-country capacity, and (v) national sovereignty in decision-making regarding GDMMs. Demonstrating and effectively communicating these elements will be as critical as ensuring their existence.

Item

Financial management practices and Agri-SME performance: does subjective financial literacy hold the key?

(Taylor & Francis Online, 2026-01-10) Mang’ana, Kulwa

this study investigates how four key financial management practices (FMps), namelythe working capital, capital budgeting, financial reporting, and financing affect agri-sMEperformance in tanzania, focusing on the mediating role of the manager’s subjectivefinancial literacy (self-perceived competence). using a cross-sectional survey of 385tanzanian agri-sME managers, data were analyzed via pls-sEM with 5,000 bootstraps.anchored in the resource Based view and behavioral finance, results show thatfinancing practices (β ≈ 0.17) and subjective literacy (β ≈ 0.13) significantly enhanceperformance (p < 0.05). Furthermore, working capital management and capital budgetingsignificantly boosted subjective literacy (p < 0.05). importantly, financial reportingsignificantly reduced perceived competence (β≈−0.23, p < 0.05), suggesting potentialcognitive misalignment. indirect paths via literacy were positive but not statisticallysignificant (p > 0.05). the study recommended that support programs must pair accessto finance with capability building that bolsters perceived financial competence, andreporting tools must be simplified.

Item

From Forest to Fuel: Exploring Uapaca Kirkiana (Wild Loquat) Extracts to Enhance Biodiesel Life

(Springer Nature, 2026-01-12) Kivevele, Thomas; Kahimbi, Henry

Biodiesel is a renewable, low-emission fuel; however, its poor oxidative stability limits its practical use. Although synthetic antioxidants like Pyrogallol can improve stability, they raise toxicity concerns, prompting interest in safer, plant-based alternatives. This study evaluates the antioxidant potential of Uapaca kirkiana extracts, specifically from fruit peels, pulp, and root bark, for enhancing the oxidation stability of biodiesel. Methanolic extraction was employed to obtain the plant extracts, and the total phenolic content (TPC) was determined using the Folin–Ciocalteu (FC) reagent. Antioxidant activity was assessed using the 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging assay, while oxidation stability of croton biodiesel was measured using the OXITEST method. Among the tested samples, fruit peel extracts exhibited the highest phenolic content and antioxidant activity. At a concentration of 1000 ppm, the fruit peel extract extended the biodiesel induction period from 3.02 h (control) to 16.25 h, compared to 14.18 h and 10.10 h for the pulp and root bark extracts, respectively. These results were benchmarked against Pyrogallol (PY), which extended the induction period to 18.30 h at the same concentration. Although PY demonstrated slightly higher effectiveness, the natural extracts, particularly from fruit peels, showed substantial potential. Notably, the fruit peel extract met and exceeded the minimum oxidation stability requirements of EN 14112 and ASTM D6751 (8 h) at just 500 ppm. These findings highlight the promise of Uapaca kirkiana extracts as eco-friendly, natural alternatives to synthetic antioxidants for improving biodiesel stability and shelf life.

Item

Enhanced multi-class object detector for bone fracture diagnosis with prescription recommendation

(Frontiers, 2026-01-12) Migayo, Daudi; Kaijage, Shubi; Swetala, Stephen; Nyambo, Devotha

Bone fractures are among the most prominent injuries in the modern world that affect all ages and races. Traditional treatment involves radiographic imaging that relies heavily on radiologists manually analyzing images. There have been efforts to develop computer-aided diagnosis tools that employ artificial intelligence and deep learning approaches. Existing literature focuses on developing tools that only detect and classify bone fractures, rather than addressing the broader issue of bone fracture management. However, evidence of scholarly works that include treatment recommendations is still lacking. Furthermore, deep learning based object detectors that achieve state-of-the-art results are computationally expensive and considered as black-box solutions. Developing countries, such as Sub-Saharan Africa, face a shortage of radiologists and orthopedists. For this reason, this paper proposes a methodological approach that uses a more efficient object detection model to diagnose long bone fractures and provide prescription recommendations. An enhanced anchoring process, known as adaptive anchoring, is proposed to improve the performance of the Regional Proposal Network and the object detection model. A Faster R-CNN model with ResNet-50/101 and ResNext-50/101 backbones was used to develop an object detection model that uses X-ray images as input. To understand and interpret the model’s decision, a Gradient-based Class Activation Mapping method was used to assess the model’s learnability. The results indicate that the proposed adaptive anchoring approach can improve computational efficiency, reducing training time by up to 29% compared to the traditional approach. Model accuracy during training and validation ranged between 94% and 98%. Overall, adaptive anchoring performed better when applied with the ResNet-101 backbone, yielding an Average Precision of 92.73%, an F1 score of 96.01%, a precision of 96.80%, and a recall of 95.23%. The study provides valuable insights into the use of computationally efficient deep learning models for medical recommendation systems. Future studies should develop models to diagnose fractures using input images from various modalities and to provide prescription recommendations.