Download and process the DRI data

For documentation and installation instructions, please see: minder_utils documentation.

Download and process the DRI data

Install anaconda, and create environment via:

# Linux
conda create --name minder_utils --file conda-linux-64.lock
# Windows
conda create --name minder_utils --file conda-win-64.lock
# Mac
conda create --name minder_utils --file conda-osx-64.lock
# M1 Mac
conda create --name minder_utils --file conda-osx-arm64.lock

Overview

NOTE: the time in the dataframe is UTC, which in the summer is 1 hour earlier then local patient time.

1. access the research portal and activate an access token
2. Copy and paste your token into Getting Started.ipynb.

Currently, the script can

1. Download the data
2. Categorize the data python main.py -formatting True. Will return an object with the following attributes

• physiological_data, the values will be averaged by date.
• activity_data
• environmental_data, the values will be averaged by date.

The weekly_loader in the 'scripts' folder supports download the activity data weekly, it can

• download all the previously collected activity data
• download the latest activity data in the last week
• put the data in a specific format
• normalise and save the data

For exmaples of installing the package, downloading, and loading the package, please see the documentation website.

Please share your ideas/code of formatting the data (activity, environmental, physiological, questionary), data visualisation or any other ideas with us. We will organise the code and share to others.

TODO

Data

1. The activity data will be aggreated hourly
2. The missing physiological data will be imputed by mean or the nearest data.
3. Textual data will be processed by text embedding.

Model

1. Unsupervised learning models including autoencoder, contrastive encoder, partial order etc.
2. Classifiers including conventional classifiers, pnn
3. NLP models for processing the textual data
4. models for data fusion