Quantifying Curricular Integration in Low-Cost Edge-AI Laboratory Instruction: A Formal Model and a Six-Week Pilot Evaluation in a Nigerian Computer Science Department

International audience Resource constraints can limit practical instruction in the Internet of Things (IoT) and embedded machine learning within undergraduate computer science programmes. This study evaluates a low-cost Edge-AI laboratory module centred on HydroSense-EI and aligned with five curriculum courses in a Nigerian university. A six-week pilot involved 30 third-year students, of whom 27 completed the pre-test and post-test assessment. Curricular integration was quantified using the proposed Integration Coupling Index (ICI), learning change was evaluated using normalised gain across six assessment domains, and affordability was examined through kit-amortisation and pedagogical cost-efficiency models. The module achieved an ICI of 0.696, corresponding to 3.07 effective courses per laboratory week relative to a single-course baseline of 1.00. Mean assessment performance increased from 20.7% to 76.8%, with a mean normalised gain of 0.711. During seven-day field deployment, the six student teams achieved mean classifier accuracy of 92.8% and mean water saving of 34.0%. The reported bill of materials was US$22.20 per team; under the stated reuse and breakage assumptions, amortised cost declined from US$4.44 per student in the first cohort to US$1.17 by the fifth. These pilot findings indicate that the proposed framework can support measurable cross-course integration and practical Edge-AI learning under constrained laboratory budgets, while requiring broader controlled and multi-site evaluation.