When Plants Respond: Electrophysiology and ML for Green Monitoring Systems

Living plants, while contributing to ecological balance and climate regulation, also function as natural sensors capable of transmitting information about their internal physiological states and surrounding conditions. This rich source of data provides potential for applications in environmental monitoring and precision agriculture. With integration into biohybrid systems, we establish novel channels of (physiological) signal flow between living plants and artificial devices. We equipped Hedera helix with a plant-wearable device called PhytoNode, to continuously record the plant’s electrophysiological activity. We deployed plants in an uncontrolled outdoor environment to map electrophysiological patterns to environmental conditions. Over five months, we collected and analyzed data using state-of-the-art and automated machine learning (AutoML). Our classification models achieve high performance, reaching macro F1-scores of up to 95% in binary tasks. AutoML approaches outperformed manual tuning and selecting subsets of statistical features further improved accuracy. Our biohybrid living system monitors the electrophysiology of plants in harsh, real-world conditions. This work advances scalable, self-sustaining, and plant-integrated living biohybrid systems for sustainable environmental monitoring.

Data repository for our paper "When Plants Respond: Electrophysiology and ML for Green Monitoring Systems", submitted to the 14th International Conference on Biomimetics and Biohybrid Systems. Please refer to the paper for more information.