Combining Disease Mechanism and Machine Learning to Predict Wheat Fusarium Head Blight

Wheat Fusarium head blight (FHB) can be effectively controlled through prediction. To address the low accuracy and poor stability of model predictions wheat FHB, a prediction method FHB that couples logistic regression mechanism-based k-nearest neighbours (KNN) is proposed in this paper. First, we selected predictive factors, including remote sensing-based meteorological factors. Then, quantitatively expressed factor weights disease occurrence development mechanisms by using model. Subsequently, integrated obtained into factors input with KNN to predict incidence FHB. Finally, generalizability models were evaluated. fields Changfeng, Dingyuan, Fengyuan, Feidong counties, Anhui Province, where often occurs, used as study area. The incidences on 29 April 10 May 2021 predicted. Compared did not consider mechanism, our increased approximately 13%. overall accuracies for two dates 0.88 0.92, F1 index was 0.86 0.94, respectively. results show made logistic-KNN had higher better than those model, thus achieving high-precision