This project was part of my Stanford Professional Certification Course in Machine Learning (xCS229ii).
The goal of the project is to detect presence of stress from heart rate measure by wrist worn PPG devices (such as fitbits, Apple watches, etc). We designed a CNN model that gave a 0.7 AUC and 0.6 accuracy. Our process is documented in the 'xCS229ii_final Paper.pdf' included in this repository.
Datasets used:
Wesad multimodal dataset: https://archive.ics.uci.edu/ml/datasets/WESAD+%28Wearable+Stress+and+Affect+Detection%29
UBFC dataset: https://ieee-dataport.org/open-access/ubfc-phys-2