Scholarlink

<p>A novel predictive model was developed for early blight <i>Alternaria solani</i>, as no solution still exists for this disease of increasing importance. Potato farming holds significant economic importance in Hainaut. Since 1987, CARAH’s late blight model has been a key decision-support tool for farmers across Hainaut and, later, the entire Walloon Region (38,000&#xa0;ha). As part of the Interreg Sytranspom project, the CARAH team developed a molecular monitoring method for <i>A. solani</i> using quantitative PCR (qPCR) on potato leaves and air samples collected throughout the growing season. Simultaneously, a predictive model for <i>A. solani</i> was developed, based on an Artificial Neural Network, integrating both historical field data and the newly established molecular monitoring approach. The model is based on a three-layer Artificial Neural Network, implemented using the KERAS package in R. It was trained on 7&#xa0;years of field observations and meteorological data from a weather station network, including solar radiation, leaf moisture, relative humidity, rainfall, and temperature. The model was successfully tested under real field conditions during a recent growing season. The results demonstrated high predictive accuracy (0.92) for both the timing of early blight onset and the severity of infections. Future work will focus on further improving the model’s performance and robustness by incorporating additional experimental data.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

New Alternaria solani Predictive Model Based on Artificial Neural Network

  • 期刊论文

  • 发表日期：

    2026年4月22日

查看全文
  • Thierry Marique,
  • Maxime Bonnave,
  • Cyrille Vryghem,
  • Benjamin Couvreur,
  • John Rivière,
  • Olivier Mahieu,
  • Deborah Lanterbecq

摘要

A novel predictive model was developed for early blight Alternaria solani, as no solution still exists for this disease of increasing importance. Potato farming holds significant economic importance in Hainaut. Since 1987, CARAH’s late blight model has been a key decision-support tool for farmers across Hainaut and, later, the entire Walloon Region (38,000 ha). As part of the Interreg Sytranspom project, the CARAH team developed a molecular monitoring method for A. solani using quantitative PCR (qPCR) on potato leaves and air samples collected throughout the growing season. Simultaneously, a predictive model for A. solani was developed, based on an Artificial Neural Network, integrating both historical field data and the newly established molecular monitoring approach. The model is based on a three-layer Artificial Neural Network, implemented using the KERAS package in R. It was trained on 7 years of field observations and meteorological data from a weather station network, including solar radiation, leaf moisture, relative humidity, rainfall, and temperature. The model was successfully tested under real field conditions during a recent growing season. The results demonstrated high predictive accuracy (0.92) for both the timing of early blight onset and the severity of infections. Future work will focus on further improving the model’s performance and robustness by incorporating additional experimental data.