Machine learning-assisted implantable plant electrophysiology microneedle sensor for plant stress monitoring

Plant electrical signals serve as a medium for long-distance signal transmission and are intricately linked to plant stress responses. High-fidelity acquisition and analysis of plant electrophysiological signals contribute to early stress identification, thereby enhancing agricultural production efficiency. While traditional plant electrophysiology monitoring methods like gel electrodes can capture electrical signals on plant surfaces, which face limitations due to the plant cuticle barrier, impacting signal accuracy. Moreover, the vast and intricate nature of plant electrical signal data, coupled with the absence of specialized large-scale models, impedes signal interpretation and plant physiological correlation. In light of these challenges, we engineered an implantable microneedle array using micromachining technology for monitoring and decoding plant electrical signals in a minimally invasive manner. This innovative sensor can securely adhere to plant tissue over extended periods, enabling the precise recording of electrical signals triggered by transient (mechanical injury) and long-term stresses (drought and salt stress). Based on the collected plant electrophysiological data, we utilized a machine learning model to analyze these signals for the early detection of plant stress with an accuracy of 99.29%. This sensor has great potential and is expected to revolutionize precision agricultural production and provide valuable help in managing plant stress more effectively.