Causality, machine learning and human insight

Martens, Harald

doi:10.1016/j.aca.2023.341585

Cited by 1 publication

(1 citation statement)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the second stage of this study, the HSR spectral data were correlated to the fifteen most significant NMR spectral regions according to the VIP plot of the PLS-DA model. Since the spectral reflectance data, such as HSR, contains overlapping signals from both the chemical composition of the sample and physical effects, such as light scattering, as well as artifacts such as illumination/shadows [64], it is usually necessary to separate the physical information from chemical information [65,66]. However, to model the additive/multiplicative effects in the sample, it is required to use the whole spectrum, which is not feasible for spectral data acquired using instruments such as drones, which are typically equipped with few wavebands, usually around six to twelve channels [67].…”

Section: Discussionmentioning

confidence: 99%

Non-Targeted Spectranomics for the Early Detection of Xylella fastidiosa Infection in Asymptomatic Olive Trees, cv. Cellina di Nardò

Ahmed,

Musio,

Todisco

et al. 2023

Molecules

View full text Add to dashboard Cite

Olive quick decline syndrome (OQDS) is a disease that has been seriously affecting olive trees in southern Italy since around 2009. During the disease, caused by Xylella fastidiosa subsp. pauca sequence type ST53 (Xf), the flow of water and nutrients within the trees is significantly compromised. Initially, infected trees may not show any symptoms, making early detection challenging. In this study, young artificially infected plants of the susceptible cultivar Cellina di Nardò were grown in a controlled environment and co-inoculated with additional xylem-inhabiting fungi. Asymptomatic leaves of olive plants at an early stage of infection were collected and analyzed using nuclear magnetic resonance (NMR), hyperspectral reflectance (HSR), and chemometrics. The application of a spectranomic approach contributed to shedding light on the relationship between the presence of specific hydrosoluble metabolites and the optical properties of both asymptomatic Xf-infected and non-infected olive leaves. Significant correlations between wavebands located in the range of 530–560 nm and 1380–1470 nm, and the following metabolites were found to be indicative of Xf infection: malic acid, fructose, sucrose, oleuropein derivatives, and formic acid. This information is the key to the development of HSR-based sensors capable of early detection of Xf infections in olive trees.

show abstract