An IoT-Driven Machine Learning Framework for Predicting Fungal Infections in Greenhouse Crops

Abstract

Fungal infections remain one of the major causes of reduced productivity and economic loss in greenhouse agriculture due to delayed detection and inadequate environmental monitoring. Conventional disease management approaches are largely reactive and depend heavily on manual inspection, resulting in late intervention and increased fungicide usage. This study focused on Internet of Things (IoT)-driven machine learning framework for predicting fungal infections in greenhouse crops using simulated IoT environmental sensing data and predictive analytics. The framework combined IoT sensors for constant acquisition of crucial greenhouse parameters such as temperature, humidity, soil moisture, and light intensity. Sensor data are transmitted through a wireless communication layer to a centralized prediction engine where machine learning algorithms are employed for disease risk classification. The three learning models, such as Random Forest (RF), Support Vector Machine, and Convolutional Neural Network (CNN), were comparatively evaluated to determine the most effective predictive approach for greenhouse fungal infection monitoring. Experimental evaluation using a synthetic greenhouse dataset demonstrated that the Random Forest model achieved the best predictive performance with an accuracy of 91.5%, outperformed SVM and CNN models in classification stability and generalization proficiency. The proposed system provides an intelligent early warning mechanism capable of supporting proactive disease management, minimizing crop losses, and improving greenhouse productivity. The study contributes to smart agriculture research by providing an integrated IoT-machine learning framework for real-time greenhouse disease prediction and decision support.