Detecting the Type and Severity of Mineral Nutrient Deficiency in Rice Plants Based on an Intelligent microRNA...

Area of Science:

• Agricultural Science
• Biotechnology
• Data Science

Background:

• Early detection of nutrient deficiencies in crops like rice is crucial for preventing yield loss.
• MicroRNAs (miRNAs) show potential as biomarkers for plant stress responses.

Purpose of the Study:

• To develop and evaluate a machine learning model for identifying the type and severity of nitrogen, phosphorus, potassium, and sulfur deficiencies in rice plants.
• To utilize plant microRNA data as input for the predictive model.

Main Methods:

• An electrochemical biosensor was employed to measure the concentration of 14 microRNAs in rice plants under nutrient deficiency.
• Machine learning models, including a genetic algorithm-optimized random forest, were used for stress prediction.
• Feature selection identified key miRNAs (miRNA167, miRNA162, miRNA169, miRNA395) for nutrient deficiency prediction.

Main Results:

• The biosensor demonstrated excellent analytical performance for microRNA quantification.
• The optimized random forest model achieved high accuracy (0.86), precision (0.94), and recall (0.87) in detecting nutrient deficiency types.
• The model accurately predicted deficiency levels with an average MSE of 0.010 and R² of 0.92.

Conclusions:

• Plant microRNA profiles, when analyzed by machine learning, can effectively identify specific nutrient deficiencies in rice.
• This approach offers a promising foundation for developing advanced plant health monitoring systems.