High throughput machine learning pipeline to characterize larval zebrafish motor behavior.

Area of Science:

• Neuroscience
• Behavioral Science
• Machine Learning

Background:

• Larval zebrafish are a valuable model for studying neural substrates of behavior due to their simple nervous system.
• Understanding zebrafish behavior is crucial for neuroscientific research.

Purpose of the Study:

• To develop and validate machine learning models for detecting and classifying larval zebrafish behaviors.
• To establish a high-throughput method for analyzing zebrafish behavioral phenotypes.

Main Methods:

• Utilized an 8 key point pose estimation model to capture zebrafish kinematics.
• Trained random forest classifiers using a semi-supervised learning framework on a manually labeled dataset.
• Validated models on spontaneous and stimulus-evoked behaviors, and drug-induced locomotor effects.

Main Results:

• Developed accurate machine learning classifiers for discreet zebrafish behaviors (stationary, scoot, turn, startle responses).
• Achieved high precision in classifying behaviors.
• Demonstrated the pipeline's utility in analyzing drug effects on locomotion.

Conclusions:

• Machine learning, particularly semi-supervised learning, provides a powerful tool for detailed behavioral phenotyping in zebrafish.
• The developed models offer a high-throughput method for studying the neural control of behavior.