https://github.com/genentech/stats4phc
Performance evaluation for the prognostic value of predictive models intended to support personalized healthcare through predictiveness curves and positive / negative predictive values.
https://github.com/genentech/stats4phc
Last synced: 4 months ago
JSON representation
Performance evaluation for the prognostic value of predictive models intended to support personalized healthcare through predictiveness curves and positive / negative predictive values.
- Host: GitHub
- URL: https://github.com/genentech/stats4phc
- Owner: Genentech
- Created: 2024-02-20T08:45:16.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2025-02-17T08:28:32.000Z (over 1 year ago)
- Last Synced: 2025-04-23T16:25:41.666Z (about 1 year ago)
- Language: R
- Homepage: https://genentech.github.io/stats4phc/
- Size: 19.9 MB
- Stars: 1
- Watchers: 4
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# stats4phc 
[](https://genentech.github.io/stats4phc/unit-test-report/)
[](https://genentech.github.io/stats4phc/)
[](https://genentech.github.io/stats4phc/coverage-report/)
Performance evaluation for the prognostic value of predictive models intended to
support personalized healthcare (phc) when the outcomes of interest are binary.
Predictiveness curves
are an insightful visualization to assess the inherent ability of such
models to provide predictions to individual patients. Cumulative versions of predictiveness
curves represent positive predictive values and 1 - negative predictive values and are also
informative if the eventual goal is to use a cutoff for clinical decision making.
In addition, predictiveness curves and their cumulative versions are naturally related to
net benefit
performance metrics to assess clinical utility for phc. Finally, some authors have
proposed a visualization that assesses both the prognostic value of predictive models and
their performance as a classifier. This package provides a variety of functions for estimation
and plotting of these performance evaluation curves and metrics.
## Installation
``` r
remotes::install_github(repo = "genentech/stats4phc")
```
For reproducibility, refer to a specific version tag, for example
``` r
remotes::install_github(repo = "genentech/stats4phc", ref = "v0.1.2")
```
## Documentation
Please refer to https://genentech.github.io/stats4phc
where you can see function reference as well as introduction vignette.
## Example
This is a basic example which demonstrates `riskProfile` function (i.e. Predictiveness curve):
``` r
library(stats4phc)
# Read in example data
auroc <- read.csv(system.file("extdata", "sample.csv", package = "stats4phc"))
rscore <- auroc$predicted_calibrated
truth <- as.numeric(auroc$actual)
# For clarity, show just PPV and 1-NPV. Or display Predictiveness curve by using "PC"
p1 <- riskProfile(outcome = truth, score = rscore, include = c("PPV", "1-NPV"))
p1$plot
# You can also access the underlying data with `p1$data`
```
and `sensSpec` function (Sensitivity and Specificity Plot):
```r
p2 <- sensSpec(outcome = truth, score = rscore)
p2$plot
```
