Abstract Search

ISEF | Projects Database | Finalist Abstract

| Print PDF

Dopavision: A Novel Supervised Machine Learning Model for Predicting Post-Cocaine Phasic Mesolimbic Dopamine Activity in Male C57BL/6J Mice

Booth Id:
CBIO074

Category:
Computational Biology and Bioinformatics

Year:
2025

Finalist Names:
Yoon, Sumin (School: Collierville High School)

Abstract:
Dysregulated dopamine in the mesolimbic system is linked to disorders like schizophrenia, depression, and addiction. Low phasic dopamine release correlates with reduced drug reward sensitivity and reduced motivation, but no reliable method exists to predict phasic dopamine levels. This study developed a prototype machine learning model, Dopavision, to predict percent changes in phasic dopamine release following common behavioral testing in mice. Using social and locomotor data, the multiple linear regression explained 60.18% of dopamine variability, while the random forest classifier algorithm achieved 83.33% accuracy. Future research can refine the model for both sexes, potentially enabling human DA prediction and early disorder prevention.

Awards Won: