Related Experiment Video
Updated: Sep 17, 2025
05:42
The Forced Swim Test as a Model of Depressive-like Behavior
Published on: March 2, 2015
37.8K
Machine learning-based model for behavioural analysis in rodents applied to the forced swim test.
Andrea Della Valle1,2, Sara De Carlo1, Gregorio Sonsini1
1School of Pharmacy, Center of Neuroscience, University of Camerino, Via Madonna delle Carceri, 62032, Camerino, MC, Italy.
Scientific Reports
|July 2, 2025
Summary
A new machine learning model accurately analyzes rodent behavior in the Forced Swim Test (FST), differentiating immobility, swimming, and climbing. This automated approach offers a standardized, unbiased tool for preclinical research on antidepressant efficacy.
Area of Science:
- Neuroscience and Behavioral Science
- Pharmacology and Drug Discovery
- Artificial Intelligence in Research
Background:
- The Forced Swim Test (FST) is a standard preclinical model for evaluating antidepressant effects and depressive-like behaviors in rodents.
- Current manual scoring of FST is labor-intensive, subjective, and prone to bias.
- Existing automated systems lack the granularity to distinguish between specific behaviors like swimming and climbing.
Purpose of the Study:
- To develop a novel, automated machine learning (ML) system for precise behavioral analysis in the FST.
- To overcome the limitations of manual scoring and current automated systems by differentiating key behavioral subtypes.
- To provide a standardized and unbiased tool for assessing antidepressant efficacy and stress responses in rodents.
Main Methods:
- Utilized a three-dimensional residual convolutional neural network (3D RCNN) to process video data directly, capturing spatiotemporal dynamics.
- Trained and validated the ML model against manual scoring in rats treated with known antidepressants (fluoxetine, desipramine).
- Further validated the model's ability to differentiate drug-induced behavioral patterns (climbing, swimming, or mixed).
Main Results:
- The ML model successfully and accurately differentiated between immobility, swimming, and climbing behaviors in rats.
- The model demonstrated efficacy in distinguishing between drugs known to elicit specific behavioral responses (amitriptyline, paroxetine, venlafaxine).
- Achieved a standardized and unbiased automated analysis of rodent behavior in the FST.
Conclusions:
- The developed 3D RCNN-based ML model offers a significant advancement for automated behavioral analysis in the FST.
- This approach provides a more accurate, efficient, and objective method for evaluating antidepressant efficacy and stress-related behaviors.
- The ML model has potential for broader application in other behavioral tests for laboratory animals.
