Key points are not available for this paper at this time.
ABSTRACT Global food security is threatened by crop diseases and nutrient deficiencies. Traditional detection methods—visual scouting, molecular diagnostics and soil testing—are reactive and only identify problems once visible symptoms appear, which often misses intervention windows. This narrative review synthesizes 173 peer‐reviewed articles (from 2012 to 2025) to critically evaluate the synergistic potential of artificial intelligence (AI) and multisensor satellite remote sensing (RS) for presymptomatic detection. We propose a four‐principal framework: (1) sensor choice must align with pathogen infection strategy; (2) detection becomes actionable when spectral deviation exceeds twice baseline noise; (3) spectral time series can estimate epidemiological parameters (e.g., latent period, Area Under the Disease Progress Curve); and (4) explainable AI (XAI) converts black‐box predictions into interpretable diagnostics. Key findings uncovered were that multispectral sensors detect biotrophic pathogens 5–10 days pre‐symptomatically via red‐edge sensitivity; hyperspectral platforms offer 7–14 days warning and that thermal sensors detect vascular wilts 1–7 days earlier. Key challenges remain, including trade‐offs between resolution and revisit frequency, atmospheric interference causing 60%–80% optical data loss in tropical regions, spectral confusion between biotic and abiotic stresses, and limited scalability for smallholder farms (< 0.5 ha). Notably, only 2 out of 11 studies used molecular validation (e.g., quantitative PCR, ELISA). Emerging solutions such as CubeSat constellations, Internet of Things (IoT)‐integrated monitoring and edge AI, could bridge the implementation gap. In conclusion, the integration of AI and RS enables a shift from reactive to presymptomatic crop health management; however, future research should prioritize direct comparative studies validated using molecular ground‐truth data.
Abdela et al. (Fri,) studied this question.