International Research Journal of Education and Technology

      Plant diseases drastically impair crop productivity and jeopardize global food security.  Automated illness prediction with machine learning (ML) provides a scalable method for early identification and management.  This research describes a hybrid machine-learning system that blends image-based deep learning (Convolutional Neural Networks) for symptom classification with traditional ML models (Random Forest) that use environmental and agronomic factors to forecast disease recurrence.  We test the method using a multi-crop dataset (leaf photos and weather recordings).  The suggested method obtains an average classification accuracy of 94.2% for image-based detection and an F1-score of 0.92 for combined illness occurrence prediction (example values).  The findings indicate that combining visual and contextual data enhances prediction robustness and facilitates prompt intervention measures.