Towards Health Status Determination and Local Weather Forecasts from Vitis vinifera Electrome.

Area of Science:

• Plant electrophysiology
• Machine learning
• Environmental monitoring

Background:

• Plant bioelectric signals may contain information beyond physiological processes.
• Recent advances in machine learning enable analysis of complex biological data.
• Vascular plants' electrical potentials are increasingly recognized for potential environmental sensing.

Purpose of the Study:

• To develop a novel framework for analyzing plant bioelectrical signals.
• To infer short-term meteorological parameters using plant electrophysiology and machine learning.
• To explore the potential of plants as bio-integrated environmental intelligence systems.

Main Methods:

• Acquired continuous bioelectrical potential data from Vitis vinifera using a multi-channel system.
• Utilized plants in various health conditions (healthy, diseased, recovering, dead).
• Trained ensemble machine learning models, including recurrent neural networks, on electrophysiological data.

Main Results:

• Machine learning models accurately predicted temperature and humidity, comparable to sensor benchmarks.
• Models showed particular skill in forecasting rapid weather transitions.
• Feature importance analysis identified plant-specific electrophysiological patterns correlated with ambient conditions.

Conclusions:

• Plant bioelectric signals can serve as a basis for ultra-local weather forecasting.
• This bioinspired approach offers passive, biologically rooted environmental monitoring.
• Findings contribute to plant-based sensing and biomimetic environmental monitoring for Earth system observation.