After a lot of hard work by the Verisian team, I’m so excited to announce the first release of variable traceability!

Previously we released code traceability on dataset level. This summarizes how datasets depend on each other from raw data to generated outputs (TLFs). This dependency graph extends across all files in a study and lets you seamlessly navigate from results to raw data inputs in seconds. Manually tracing a result in our demo study required a programmer to go through eight different files. With dataset traceability, we can now automatically assemble a virtual file containing only the 1,010 lines of code required to create the dataset of interest:

Datasets contain many variables, but often you are only interested in a subset or even a single variable. For example, statistical tests often rely on a small set of variables that you want to understand individually. With variable traceability, you can further condense the 1,010 lines of code down to only those required for the derivation of a single variable. This results in the most concise representation (often 20-80 lines of code) that are easy to read and understand, as the traces exclude any distracting code related to other derivations (Figure 2 & 3).

Using Verisian’s dataset traceability in a Phase 3 trial, a programmer can save over 90% of the time required to ensure specification adherence and handle internal and external inquiries. Crucially, these time savings are expected to increase significantly with the implementation of variable traceability compared to dataset traceability.

Variable traceability is the holy grail in code tracing: select any variable of interest (VOI) and you will receive:

1. A variable graph, illustrating your VOI’s upstream variable dependencies and downstream effects
2. Only the lines of code that contribute to its derivation and downstream effects

Variable traceability allows you to see everything necessary to understand a variable without any distracting information.

‍