Python data analysis and models examining an academic clinical depression medication study. The data was compiled by multiple academic institutions and tracked depression patients Hamilton scores over 11 therapy sessions. The participants were either given an SSRI or placebo.
The goal was to use various classifiers to determine if a depression patient was responsive to SSRI treatment as early as possible. My results show that by the 4th week there is high accuracy in prediction, which provides a 50% decrease in diagnosis time compared to the standard 8th week determination.
This Jupyter Notebook contains a running narrative of data exploration, cleaning, and machine learning model selection/optimization.