Tool package for epidemiologic analyses
Project description
zEpid
zEpid is an epidemiology analysis package, providing easy to use tools for epidemiologists coding in Python 3.5+. The purpose of this library is to provide a toolset to make epidemiology ez. A variety of calculations and plots can be generated through various functions. For a sample walkthrough of what this library is capable of, please look at the tutorials available at https://github.com/pzivich/PythonforEpidemiologists
A few highlights: basic epidemiology calculations, easily create functional form assessment plots, easily create effect measure plots, and causal inference tools. Implemented estimators include; inverse probability of treatment weights, inverse probability of censoring weights, inverse probabilitiy of missing weights, augmented inverse probability of treatment weights, timefixed gformula, Monte Carlo gformula, Iterative conditional gformula, and targeted maximum likelihood (TMLE). Additionally, generalizability/transportability tools are available including; inverse probability of sampling weights, gtransport formula, and doubly robust generalizability/transportability formulas.
If you have any requests for items to be included, please contact me and I will work on adding any requested features. You can contact me either through GitHub (https://github.com/pzivich), email (gmail: zepidpy), or twitter (@zepidpy).
Installation
Installing:
You can install zEpid using pip install zepid
Dependencies:
pandas >= 0.18.0, numpy, statsmodels >= 0.7.0, matplotlib >= 2.0, scipy, tabulate
Module Features
Measures
Calculate measures directly from a pandas dataframe object. Implemented measures include; risk ratio, risk difference, odds ratio, incidence rate ratio, incidence rate difference, number needed to treat, sensitivity, specificity, population attributable fraction, attributable community risk
Measures can be directly calculated from a pandas DataFrame object or using summary data.
Other handy features include; splines, Table 1 generator, interaction contrast, interaction contrast ratio, positive predictive value, negative predictive value, screening cost analyzer, counternull pvalues, convert odds to proportions, convert proportions to odds
For guided tutorials with Jupyter Notebooks: https://github.com/pzivich/PythonforEpidemiologists/blob/master/3_Epidemiology_Analysis/a_basics/1_basic_measures.ipynb
Graphics
Uses matplotlib in the background to generate some useful plots. Implemented plots include; functional form assessment (with statsmodels output), pvalue function plots, spaghetti plot, effect measure plot (forest plot), receiveroperator curve, dynamic risk plots, and L'Abbe plots
For examples see: http://zepid.readthedocs.io/en/latest/Graphics.html
Causal
The causal branch includes various estimators for causal inference with observational data. Details on currently implemented estimators are below:
GComputation Algorithm
Current implementation includes; timefixed exposure gformula, Monte Carlo gformula, and iterative conditional gformula
Inverse Probability Weights
Current implementation includes; IP Treatment W, IP Censoring W, IP Missing W. Diagnostics are also available for IPTW. IPMW supports monotone missing data
Augmented Inverse Probability Weights
Current implementation includes the augmentedIPTW estimator described by Funk et al 2011 AJE
Targeted Maximum Likelihood Estimator
TMLE can be estimated through standard logistic regression model, or through userinput functions. Alternatively, users
can input machine learning algorithms to estimate probabilities. Supported machine learning algorithms include sklearn
Generalizability / Transportability
For generalizing results or transporting to a different target population, several estimators are available. These include inverse probability of sampling weights, gtransport formula, and doubly robust formulas
Tutorials for the usage of these estimators are available at: https://github.com/pzivich/PythonforEpidemiologists/tree/master/3_Epidemiology_Analysis/c_causal_inference
Gestimation of Structural Nested Mean Models
Single timepoint gestimation of structural nested mean models are supported.
Sensitivity Analyses
Includes trapezoidal distribution generator, corrected Risk Ratio
Tutorials are available at: https://github.com/pzivich/PythonforEpidemiologists/tree/master/3_Epidemiology_Analysis/d_sensitivity_analyses
Project details
Release history Release notifications  RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Filename, size  File type  Python version  Upload date  Hashes 

Filename, size zepid0.9.0py2.py3noneany.whl (555.7 kB)  File type Wheel  Python version py2.py3  Upload date  Hashes View 
Filename, size zepid0.9.0.tar.gz (535.7 kB)  File type Source  Python version None  Upload date  Hashes View 
Hashes for zepid0.9.0py2.py3noneany.whl
Algorithm  Hash digest  

SHA256  fd96cbe4df5240b3363a16e5f6da0471e0bb2bd4f335d1c097b470c6498994cb 

MD5  cfe0f582281f8c34d75bc86e0417480c 

BLAKE2256  abfa4ec88f3b159155c93697140de2352e2721ad3505dd45064c3d0a00a1c285 