For documentation and installation instructions, please see: minder_utils documentation.
Install anaconda, and create environment via:
conda env create -f environment.yml
NOTE: the time in the dataframe is UTC, which in the summer is 1 hour earlier then local patient time.
- access the research portal and activate an access token
- Copy and paste your token into Getting Started.ipynb.
Currently, the script can
- Download the data
- 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
Please check the Instruction.ipynb for usage.
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.
Data
- The activity data will be aggreated hourly
- The missing physiological data will be imputed by mean or the nearest data.
- Textual data will be processed by text embedding.
Model
- Unsupervised learning models including autoencoder, contrastive encoder, partial order etc.
- Classifiers including conventional classifiers, pnn
- NLP models for processing the textual data
- models for data fusion