Dataset Overview

This repository contains a dataset from PhysioNet: VitalDB, a high-fidelity multi-parameter vital signs database in surgical patients.

Data Features:

• Patient demographics
• Medical history
• Surgical information: anesthesia and operation time, surgical approach
• Physiological data.

This repository provides tools for preprocessing and analyzing this data, focusing on handling missing values, encoding categorical variables, and scaling numerical features.

Python

---

Installation

• Create environment for the required dependencies

conda create -n pyphysionet python~=3.9.0
conda activate pyphysionet
cd [CLONED_DIRECTORY]
pip install -r requirements.txt

• Buildup the src path

conda install conda-build
conda develop src
cd src

Pre-processing

• Identify patients with abnormal_data

from src.clinical_data import abnormal_data

• Filter patients by abnormal lab results

from src.clinical_data import select_data

• Filter patients by ASA level

from src.clinical_data import select_asa

• View patient-specific abnormal data

from src.clinical_data import select_pt

• Identify patients with pre-op diabetes (DM) or hypertension (HTN)

from src.clinical_data import medical_history

• Calculate average anes and op time.

from src.clinical_data import anes_op_time

Statistical Analysis

Chi-square analysis

Perform chi-square tests to check if two categorical variables are related or independent.

example usage

from statistics import chi2

chi2(df,
     dependent_var='death_inhosp',
     independent_var=['preop_htn', 'sex']
     )

Data visualisation

example usage

from visulisaion import figure_by_gender

figure_by_gender(df, 
                 bar_chart=False, # to create a bar chart if it is True
                 pie_chart=True, # to create a pie chart if it is True
                 filter_col='opname', # any column to filter on
                 filter_val='Lung lobectomy') # any value to filter on 

SAS

---

Installation

Using SAS OnDemand for Academics

Usage

• libname_surg.sas: To create a library named surg
• csv_to_sas.sas: To import CSV file into a SAS dataset
• clinical_data.sas: To standardise variable name and mapping raw data to CDISC SDTM (Study Data Tabulation Model), including Demographics, Adverse Events, and Laboratory Results
• ouput_path.sas: To create a macro variable for output
• exploring_data.sas: To create Frequency Report, Plots

Contact

Wei Jan Chang, [email protected]