EUVIMEDEUVIMED entdecken
Crossref

When the Screening Threshold Does Not Depend on the Epidemic: Separability in Edge-Assisted Crop Disease Surveillance

Michael Muigai, Peterson Nyaga, Morris Kimathi Mwangi, Samuel Macharia Kang'iri

Journal of Applied Science, Engineering and Technology for Development · 16.11.2026 · 6 (1)

DOI: 10.33803/yrtpzm.2 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Applied Science, Engineering and Technology for Development, online erschienen am 16.11.2026. Verfasst von Michael Muigai und 3 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

A camera node that screens leaf images locally and offloads only the suspicious ones over a 2G link has two design variables, the screening threshold and the inspection interval. They are usually chosen separately, although disease spread between inspections links them, and inspection design theory recommends optimizing them together. Whether that is needed for a screening node has not been examined. This paper composes the joint objective from a likelihood ratio test on a scalar leaf index, a renewal model of the uplink and a logistic disease progress curve. It shows that at an interior optimum the threshold depends only on the class-conditional densities of the index and the recall of the cloud classifier, while the infection rate, the link, the payload, and the cost weights set the interval alone. The proof needs three properties: a detection probability per inspection that does not change with time since onset or between leaves, a loss that depends on the mean detection delay, and interval-dependent costs that depend on the offload rate. Numerically the threshold stays at 0.2835 to within $1.4\times 10^{-6}$ over the swept parameters and four loss functions, while the optimal interval ranges from 5.7 to 22.2 days. Relaxing each property, the threshold is unchanged under a nonlinear tariff on daily volume and under an energy budget, moves to between 0.291 and 0.326 when recall rises with severity, and moves most to between 0.239 and 0.271 when parts of the plot differ in both infection rate and leaf index distribution. There the threshold computed from pooled data costs up to 8.5% more than the optimum. Where the properties hold, the threshold can be set once for every site, and joint optimization saves only 4.9% against a weekly schedule. All parameter values are assumed or taken from datasheets.

Zitieren

APA

Muigai, M., Nyaga, P., Mwangi, M. K., & Kang'iri, S. M. (2026). When the Screening Threshold Does Not Depend on the Epidemic: Separability in Edge-Assisted Crop Disease Surveillance. Journal of Applied Science, Engineering and Technology for Development, 6(1). https://doi.org/10.33803/yrtpzm.2

Vancouver

Muigai M, Nyaga P, Mwangi MK, Kang'iri SM. When the Screening Threshold Does Not Depend on the Epidemic: Separability in Edge-Assisted Crop Disease Surveillance. Journal of Applied Science, Engineering and Technology for Development. 2026;6(1). doi:10.33803/yrtpzm.2

BibTeX
@article{Muigai2026,
  author = {Muigai, Michael and Nyaga, Peterson and Mwangi, Morris Kimathi and Kang'iri, Samuel Macharia},
  title = {When the Screening Threshold Does Not Depend on the Epidemic: Separability in Edge-Assisted Crop Disease Surveillance},
  journal = {Journal of Applied Science, Engineering and Technology for Development},
  year = {2026},
  volume = {6},
  number = {1},
  doi = {10.33803/yrtpzm.2}
}
Kostenlos starten →