A Wrapper for methods of combining multiple tests

A custom implementation of all state of art methods for combining multiple tests, written from scratch in Python 3, API inspired by SciKit-learn. This module implements 8 methods:

• PEPE - Combining diagnostic test results to increase accuracy -
• SULIU - Linear Combinations of Multiple Diagnostic Markers -
• LOGISTIC
• STEPWISE - Linear combination methods to improve diagnostic/prognostic accuracy on future observations -
• MIN_MAX - A min-max combination of biomarkers to improve diagnostic accuracy -
• RANDOM FOREST
• SVMl
• SVMr

Features

R and Python friendly

Usage examples

Python

from numpy.random import multivariate_normal
from utility.py import *. # download "utility.py" in the working directory.
u0 = [0.1,0.1,0.1, 0.1]; u1 = [0.6, 0.8, 1, 1.2]
sigma = [[1,0.5,0.5,0.5],
          [0.5,1,0.5,0.5],
          [0.5,0.5,1,0.5],
          [0.5,0.5,0.5,1]]
X0 = multivariate_normal(u0, sigma, size = 100)
X1 = multivariate_normal(u1, sigma, size = 80)
model = AllMethod(method= 'SULIU', bool_trans= False).fit(X0_train,X1_train)
_,_, auc = model.predict(X0_val,X1_val)
model.roc_plot(X0,X1)

R

library(reticulate)
source_python("utility.py") ## download "utility.py" in the working directory.
mod = AllMethod(method= 'logistic', bool_trans = TRUE);
mod$fit(X0_train,X1_train); 
mod$predict(X0_val,X1_val)
mod$roc_plot(X0_val,X1_val)

Built With

• Dropwizard - scikit-learn API
• Maven - Dependency Management
• ROME - Used to generate RSS Feeds

Versioning

We use SemVer for versioning. For the versions available, see the tags on this repository.

Authors

• Chengning Zhang - Initial work - Statistical methods for combining multiple diagnostic tests: a review

License

This project is licensed under the MIT License - see the LICENSE.md file for details