Automated analysis of feeding dynamics from electromyographic recordings in a blood-sucking insect

Abstract

Feeding behavior in blood-sucking insects relies on gustatory evaluation to decide on sustained ingestion, yet quantifying this process from electromyogram (EMG) recordings is labor-intensive. Here we developed MyoRec, an automated computational framework employing machine learning to analyse EMG signals from the triatomine bug Rhodnius prolixus. Using recordings under appetitive and aversive conditions, a convolutional neural network detected ingestion events with 97.7% accuracy. Automated analysis revealed distinct feeding dynamics, with prolonged ingestion and higher pumping frequency under appetitive stimuli, compared to rapid feeding cessation under aversive stimuli. MyoRec substantially reduces analysis time while maintaining accuracy, providing a scalable tool to investigate how gustatory cues modulate feeding decisions in hematophagous insects.