The VR Eye-tracking Cognitive Assessment (VECA): A Portable and Efficient Dementia Screening Tool Using Eye-Tracking Technology, Machine Learning, and Virtual Reality
Programs of building and evaluating models mentioned in this paper were developed using Python 3.8.
# install other required packages
pip install -r requirements.txt
Unreal eye tracking series and the corresponding info data were randomly generated in data folder only to run the code.
Results in the paper could be reproduced only via original patient data. Non-identifiable patient data are available upon request to the corresponding author.
This repository provides all core components of the pipeline of eye tracking data processing, feature extraction, modeling and evaluation based on the framework of VECA. The components or configuration such as /src/utils/eyeMovement.py and /src/config/settings.py could be customized for related research.
# train a specific model
python train.py -m gbrt
# train a specific model and output feature importance
python train.py -m svr --importance
# train a specific model and output shap value analysis
python train.py -m gbrt --shap# check all options
python train.py -h
usage: train.py [-h] [-m {svr,mlp,gbrt,lasso}] [--info_dir INFO_DIR] [--log_dir LOG_DIR] [--model_dir MODEL_DIR] [--importance] [--shap] [--roc]
optional arguments:
-h, --help show this help message and exit
-m {svr,mlp,gbrt,lasso}, --model {svr,mlp,gbrt,lasso}
Specify which ML model to train and evaluate.
--info_dir INFO_DIR Specify file path of data info excel of training data.
--log_dir LOG_DIR Specify directory path of log files.
--model_dir MODEL_DIR
Specify model persistent directory.
--importance Specify whether to output feature importances.
--shap Specify whether to compute normalized shap values.
--roc Specify whether to analyze education grouped classification ROCs.