HKU Diabetes Data Analytics

hku_diabetes is a data analytics module tailored made for analysing diabetes clinical data collected at Queen Mary Hospital of the Hong Kong University.

hku_diabetes is compatible with: Python 3.5-3.7.

Installation

If you don't have Python installed already, you could download Python 3.7 directly here.

You can obtain hku_diabetes either by download a zip by clicking the 'Clone or download' button from this page, or simply put the following in your terminal (cmd) if you have git.

git clone https://github.com/luithw/hku_diabetes.git

You then need to install all the required packages by pip.

pip install -r requirements.txt

Alternatively, if you just want to use the package, you can install the package itself directly with pip.

pip install hku_diabetes

Perform data analytics

The package expects to find the data under a directory (folder) named raw_data in the working directory. Please contact Emmanuel Wong at the Department of Medicine to obtain the data.

Once you have put the data in the working directory. First test everything is OK by executing the following command in your terminal. It execute all the functions with test configuration on a small number of data to make sure there is no error.

python main.py

The first time it is executed, it loads the raw data in inefficient html format and saves them as CSV in the processed_data directory, so that it loads faster the next time.

If the above command is executed without error. Simply add run after the previous command. It will then execute on the entire dataset.

python main.py run

Use the package in your script

First import the diabetes data using the import_all function.

from hku_diabetes.importer import import_all

data = import_all()

The core analytics logic is available in the Analyser class.

from hku_diabetes.analytics import Analyser

analyser = Analyser()

The run method of Analyser takes the data and compute the results. It also saves the results in the output/plot directory automatically.

results = analyser.run(data)

You can plot all the data and results using the plot_all function.

from hku_diabetes.plot import plot_all

plot_all(analyser)

To load the previous analysed results without running the whole analytics again, simply call the load method of the Analyser.

results = analyser.load()

You can change the configuration of inport_all and Analyser by extending the DefaultConfig class.

from hku_diabetes.config import DefaultConfig

class MyConfig(DefaultConfig):
    required_resources = ["Creatinine", "Hba1C", "Medication"]	
    ckd_thresholds = (15, 30, 45, 60, 90)
    min_analysis_samples = 10
    eGFR_low_pass = "90d" 

data = import_all(config=MyConfig)
results = analyser.run(data, config=MyConfig)

For a full list of available configuration options and functions, please see the documentation.

Support

For any quires, you can send me an email to luithw@hku.hk.

Enjoy!