Home

The goal of the Analysis Toolbox platform is to provide a mix of R script templates for a variety of common data analysis methods in biomedical research, as well as providing documentation on when to use these methods and how to interpret them. Put simply the toolbox is composed by an input-algorithm-output structure, where the input is about the situations when to use a certain method, the algorithm is the method itself represented by the R script, and the output is represented by documentation on how to interpret the results from these methods.

Algorithm

Our goal is to create a series of demand-driven templates that will likely fall into the following areas:

1. Cleaning templates that are database-specific, such as for UNOS, NIS, SEER-Medicare, among others
2. Table templates for socio-demographic info and different model types
3. Graphic templates for the most commonly used graphics for each type of model, including survival, glm, among others
4. Templates for each of the most common designs: clinical trials, secondary data analyses including propensity scores, scales, among others
5. Templates to deal with big data, automating and hiding the complexity behind operations such as communication with large database systems, parallel computing, linked open data, natural language processing, among others

Input

The input section for each method will contain the type of data required to run a certain analysis. For example, a two-sample unpaired t-test requires a continuous variable as its outcome, and a dichotomous variable as its predictor. The goal of explaining the input side of a method to improve the way clinical researchers communicate with statisticians and data scientists.

Output

The output section displays the output of R commands, graphics that might be associated with that methods, and tables representing typical output for that method that would be included in a peer-reviewed publication. This output goes along extensive comments on how to interpret each aspect of the output, so that clinical researchers can interpret results in relation to their clinical knowledge.

Annotated references

The last section of the Analysis Toolbox Platform is to provide an annotated list of bibliographic references that explain the method in a way that goes from the simplest to the more detailed and technical.