https://github.com/emma-wilson/in-vitro-screening

Open data and code for published paper.
https://github.com/emma-wilson/in-vitro-screening

machine-learning open-code open-data screening

Last synced: 4 days ago
JSON representation

Open data and code for published paper.

• Host: GitHub
• URL: https://github.com/emma-wilson/in-vitro-screening
• Owner: emma-wilson
• Created: 2022-07-14T11:59:52.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2023-02-07T18:06:55.000Z (almost 2 years ago)
• Last Synced: 2024-11-20T11:04:06.897Z (2 months ago)
• Topics: machine-learning, open-code, open-data, screening
• Language: R
• Homepage: https://doi.org/10.1042/CS20220594
• Size: 4.32 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# In Vitro Screening Comparison

Data and code to accompany the paper ["Screening for in vitro systematic reviews: a comparison of screening methods and training of a machine learning classifier"](https://doi.org/10.1042/CS20220594) published in *Clinical Science*.

## Code scripts

All code is written in R using R Markdown documents. R version number and package version numbers are included in each script. Details of each script are below. Run each script in order.

- **1-1_data cleaning.Rmd:** Prepare screening method comparison data for analysis; remove excluded data

- **1-2_calculate_performance.Rmd:** Calaculate the performace (sensitivity and specificity) of screening methods at various thresholds

- **1-3_analyse_performance.Rmd:** Plot the performance in a ROC curve and determine the optimal threshold for regex screening methods

- **1-4_search_term_retrieval_comparison.Rmd:** Additional analysis comparing retrieval of studies using the planned vs actual search terms

## Data

Data are stored in the following folders:

- **data-raw:** raw data to be processed

- **data:** data which have undergone some processing

- **data-analysis:** final clean datasets

- **data-ml_input:** input data required to train ML

- **data-ml_output:** output data from ML

## Functions

Machine learning (ML) functions are in the `functions` folder. Please not that the information required to configure the ML API are **not** included as we do not have permission to share this.

## Plots

Plot outputs (in PDF file format) are in the `figures` folder.

- **regex_histogram.pdf:** histograms showing number of regex matches against (a) tiab and (b) full text

- **screening_roc.pdf:** figure from screening comparison part of project

- **ml_roc.pdf:** figure from machine learning part of project