The continued advancements in Internet of Things (IoT) and Machine Learning (ML) technologies have led to their adoption in various domains including in industries for predictive maintenance among other applications. Given the resource constraints of IoT devices, they cannot process the resource-intensive ML algorithms hence data collected by the devices are first sent to the cloud where the algorithms are hosted for processing and inference with the results being sent back to the devices for action and/or notifications. The need to transmit data to the cloud for processing leads to increased costs, energy consumption, and high latencies affecting the implementation of the solution. Interestingly with Tiny Machine Learning (TinyML), it is possible to develop algorithms enabling edge inference on resource-constrained devices. From existing review papers, the researchers were not able to find, a comprehensive review with a focus on this area showing the need for a targeted review that can shed light on how TinyML can be tailored for predictive maintenance tasks in industries. This study therefore presents a systematic literature review of the application of TinyML in predictive maintenance in industrial settings. TinyML overview and its benefits are presented, a TinyML process flow is proposed and various use cases and their classifications have been presented. Through this exploration, the study shows the critical need for TinyML-driven solutions in predictive maintenance, identifies the existing challenges, and proposes a roadmap for future research.

The activities of birds present increasing challenges in agriculture, aviation, and environmental conservation. This has led to economic losses, safety risks, and ecological imbalances. Attempts have been made to address the problem, with traditional deterrent methods proving to be labour-intensive, environmentally unfriendly, and ineffective over time. Advances in artificial intelligence (AI) and the Internet of Things (IoT) present opportunities for enabling automated real-time bird detection and repellence. This study reviews recent developments (2020–2025) in AI-driven bird detection and repellence systems, emphasising the integration of image, audio, and multi-sensor data in IoT and edge-based environments. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses framework was used, with 267 studies initially identified and screened from key scientific databases. A total of 154 studies met the inclusion criteria and were analysed. The findings show the increasing use of convolutional neural networks (CNNs), YOLO variants, and MobileNet in visual detection, and the growing use of lightweight audio-based models such as BirdNET, MFCC-based CNNs, and TinyML frameworks for microcontroller deployment. Multi-sensor fusion is proposed to improve detection accuracy in diverse environments. Repellence strategies include sound-based deterrents, visual deterrents, predator-mimicking visuals, and adaptive AI-integrated systems. Deployment success depends on edge compatibility, power efficiency, and dataset quality. The limitations of current studies include species-specific detection challenges, data scarcity, environmental changes, and energy constraints. Future research should focus on tiny and lightweight AI models, standardised multi-modal datasets, and intelligent, behaviour-aware deterrence mechanisms suitable for precision agriculture and ecological monitoring.

According to a 2020 report from the World Health Organization (WHO), household air pollution has led to over 3 million deaths globally, with recent statistics showing a worsening situation in Africa. Integrating Internet of Things (IoT) and Artificial Intelligence (AI) technologies can help address this global challenge. IoT enables real-time data collection for monitoring pollution levels, while AI algorithms predict pollution before it reaches hazardous levels. However, existing solutions are not tailored to the African context, where wood fuel is a primary pollutant, and they predominantly focus on monitoring rather than prediction. This study presents the design and implementation of an IoT-based solution for monitoring and predicting indoor air pollution in rural African households. The system collects data in real time and transmits it to the cloud for storage, processing, and analysis, with alerts to users when pollution is detected. An AI model was successfully trained and tested to predict indoor air pollution based on the collected data. The results indicate that this approach significantly improves the accuracy and timeliness of pollution alerts, potentially reducing health risks associated with indoor air pollution. The successful implementation and testing of the system demonstrate its potential for broader applications in various indoor environments.

Background: Chronic diseases such as diabetes, hypertension, and cardiovascular conditions continue to burden African public health systems, especially due to late diagnosis. This study explores the application of Machine Learning (ML) for the early detection of diabetes using a localized dataset of 768 electronic health records from a clinic in Africa. Design and methods: A Design Science Research methodology was used to evaluate and compare different ML algorithms which includedDecision Trees, Support Vector Machines, Naïve Bayes, and a Neural Network (NN). preprocessing and hyperparameter tuning was applied to optimized the model perfomance. The models were tested for feasibility in edge-based deployment scenarios which are ideal for implimentation in the African setting. Results: The optimized NN model achieved the highest accuracy (89%), minimal latency (1 ms), and low memory usage (1 kB RAM), making it suitable for deployment in resource-constrained environments. While the dataset is limited in scope, it sets a foundation for future cross-regional studies. Conclusion: This study demonstrates the potential of edge-deployable ML models in supporting early chronic disease detection in Africa and recommends future work in regulatory alignment, ethical safeguards, and multi-site validations.

Ensuring access to safe drinking water remains a major challenge in developing regions, where traditional water quality monitoring methods are often manual, time-consuming, and unable to provide real-time insights. These limitations delay detection of contamination and hinder timely intervention. The proposed model integrates low-cost sensors to measure pH, turbidity, temperature, and total dissolved solids (TDS), along with an embedded microcontroller for real-time data acquisition and processing. A four-layer architecture consisting of sensing, edge processing, GSM-based communication, and cloud visualisation was implemented. A mixed-methods approach was used, combining quantitative performance evaluation with qualitative stakeholder feedback. The model demonstrated reliable real-time monitoring with minimal measurement deviations (±0.1 pH, ±0.3 NTU, ±10 ppm TDS, ±0.2°C). Stable data transmission was maintained despite intermittent connectivity. Stakeholders reported improved usability, faster decision-making, and high confidence in system outputs. The results highlight the feasibility of deploying low-cost IoT devices for continuous water quality monitoring in resource-constrained environments.