Environmental control in crop production typically relies on real–time sensor data to monitor environmental conditions. However, the data does not reflect the immediate physiological status of crops, limiting the precision of crop-based environmental management. To enhance crop productivity, quality, and resource use efficiency, it is necessary to collect direct information about crop health status and reflect this in crop–specific environmental control. Plant electrical signals, which reflect physiological responses to external environment, can be used as biosensors for crop health. Here, electrical signals from lettuce seedlings were measured under different temperature and lighting conditions. Machine learning techniques were applied to both raw and extracted feature signals to classify the electrical responses to each environmental parameter. Lettuce seedlings were exposed to different air temperature conditions, with and without light. The electrical signals were detected through electrode needles inserted into the stems of the seedlings under each condition. Gaussian smoothing, moving average, and peak–to–peak methods were used to remove noise from raw electrical signals and extract the features, which were then used in a machine learning model to classify the temperature and light environments. The results showed that decision tree model achieved high F1 scores of 0.9778 for temperature and 0.9316 for lighting conditions. Additionally, evaluating the generalisation performance of the model by adding another lettuce variety resulted in F1 scores confirmed the robustness of model. These findings suggest that plant electrical signals can be used to monitor crop health in complex environments, offering the potential for more precise environmental management in crop production.