Statistical analysis for patient's records

This series of notebooks provide a collection of algorithms that facilitate an end user to leverage from the already processed data.

Basic operation

Neonatal Maternal Hospital's patient records software outputs several kind of data in various formats. Specifically for this project I choose for it to be .cvs files, in order to be able to work easily with pandas and dataframes.

It has three main modules:

• Emergency
• Hospitalization
• Ambulatory

For each module, treatment is very similar, though it differs in some sutil details. Roughly the analysis made can be summed up as follows:

• preprocess - 'cleans' and formats original .csv after converting it in a pandas dataframe, in order to be processed later on
• concatenate - allows to 'merge' serveral .csv files, in case output is monthly and a year anlaysis is required.
• medical care - number of incomes by patient, responding to various criteria:
  • total
  • by hour of the day
  • by day of the week
  • by age group
• top 20 professionals with most interventions
• top 20 most coded diagnostics
• reason for discharge
• length of stay - averages in patients length of stay

Note: for discretion reasons, all content in the .cvs files has been modified to fulfil this notebook showcase, and reduce the risk of the potential privacy leakage.

Streamlit implementation

Lastly, this series of algorithms was bulked in a website leveraging python's streamlit library.