{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,20]],"date-time":"2026-07-20T06:02:33Z","timestamp":1784527353468,"version":"3.55.0"},"posted":{"date-parts":[[2026,7,19]]},"reference-count":0,"publisher":"American Chemical Society (ACS)","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>The widespread contamination of wheat with deoxynivalenol (DON) necessitates rapid, non-destructive screening methods. Although near-infrared (NIR) spectroscopy offers a promising cost-effective solution, indirect detection via Fusarium-induced sample compositional changes often suffers from high model variance and overfitting. This study systematically evaluated the impact of seven methodological factors, namely dataset selection, dataset balancing, sample grinding, toxin type, sample weighting, spectral preprocessing, and variable selection, using model population analysis (MPA) to assess the performance of partial least squares discriminant analysis (PLS-DA) sub-models for classifying wheat samples as compliant or non-compliant with the European Commission (EC) limit of 1000 \u03bcg\/kg DON. By using repeated random splits with a stratification constraint across the seven methodological factors, 12,800 PLS-DA sub-models were generated. To study the effects of each factor, a generalized linear mixed model (GLMM) was employed. The results indicated that model performance was highly influenced by all methodological factors, with the toxin type and dataset balancing showing the greatest impacts on the F1-score and True Positive Rate (TPR), respectively. The optimal pipeline achieved an estimated F1-score of 0.74 \u00b1 0.01 and a TPR of 0.84 \u00b1 0.01 by utilizing Austrian-only wheat samples, intact kernels, standard normal variate (SNV) preprocessing, and downsampling for data balancing, without variable selection. Notably, samples labelling based on total DON (tDON, sum of DON and its acetylated and modified forms) improved classification performance, indicating that there might be a stronger correlation with Fusarium-induced sample compositional changes with tDON than with DON. Furthermore, applying an observation-weighting approach to highly uncertain samples mitigated the propagation of reference analytical error, significantly reducing false negative misclassifications near the regulatory threshold. Overall, our results provide a rigorously validated framework for optimizing complex chemometric pipelines, establishing a robust NIR methodology for challenging application such as indirect non-destructive screening of DON in wheat.<\/jats:p>","DOI":"10.26434\/chemrxiv.15006267\/v1","type":"posted-content","created":{"date-parts":[[2026,7,20]],"date-time":"2026-07-20T05:06:46Z","timestamp":1784524006000},"source":"Crossref","is-referenced-by-count":0,"title":["Model Population Analysis to Optimize Near-Infrared-Based Screening of Deoxynivalenol in Wheat"],"prefix":"10.26434","author":[{"ORCID":"https:\/\/orcid.org\/0009-0000-7017-2746","authenticated-orcid":false,"given":"Dedy L.","family":"Nadeak","sequence":"first","affiliation":[{"name":"Institute of Bioanalytics and Agro-metabomics","place":["Tulln an der Donau, Austria"]},{"name":"BOKU University","place":["Tulln an der Donau, Austria"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-6195-1319","authenticated-orcid":false,"given":"Patrick","family":"Rennhofer","sequence":"additional","affiliation":[{"name":"Institute of Bioanalytics and Agro-metabomics","place":["Tulln an der Donau, Austria"]},{"name":"BOKU University","place":["Tulln an der Donau, Austria"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Maximilian","family":"Anlanger","sequence":"additional","affiliation":[{"name":"Institute of Bioanalytics and Agro-metabomics","place":["Tulln an der Donau, Austria"]},{"name":"BOKU University","place":["Tulln an der Donau, Austria"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3302-0732","authenticated-orcid":false,"given":"Michael","family":"Sulyok","sequence":"additional","affiliation":[{"name":"Institute of Bioanalytics and Agro-metabomics","place":["Tulln an der Donau, Austria"]},{"name":"BOKU University","place":["Tulln an der Donau, Austria"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6843-9755","authenticated-orcid":false,"given":"Rudolf","family":"Krska","sequence":"additional","affiliation":[{"name":"Institute of Bioanalytics and Agro-metabomics","place":["Tulln an der Donau, Austria"]},{"name":"BOKU University","place":["Tulln an der Donau, Austria"]},{"name":"Institute for Global Food Security, School of Biological Sciences, Queen's University Belfast","place":["Belfast, UK"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0179-2217","authenticated-orcid":false,"given":"Stephan","family":"Freitag","sequence":"additional","affiliation":[{"name":"Institute of Bioanalytics and Agro-metabomics","place":["Tulln an der Donau, Austria"]},{"name":"BOKU University","place":["Tulln an der Donau, Austria"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"316","container-title":[],"original-title":[],"link":[{"URL":"https:\/\/chemrxiv.org\/doi\/pdf\/10.26434\/chemrxiv.15006267\/v1","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,20]],"date-time":"2026-07-20T05:06:46Z","timestamp":1784524006000},"score":1,"resource":{"primary":{"URL":"https:\/\/chemrxiv.org\/doi\/full\/10.26434\/chemrxiv.15006267\/v1"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,19]]},"references-count":0,"URL":"https:\/\/doi.org\/10.26434\/chemrxiv.15006267\/v1","relation":{},"subject":[],"published":{"date-parts":[[2026,7,19]]},"subtype":"preprint"}}