# 1.0 release

After almost 1.5 years of development, we are finally releasing mobgap 1.0.
While core functionality of the library remains unchanged since the early alpha, we spent the last 8 months ensuring that all algorithms work as expected and match the scientific performance thresholds we decided on as part of the Mobilise-D project.

For this we developed a fully reproducible benchmarking framework that allowed us to calculate all relevant performance metrics of all newly implemented algorithms and compare them with the old implementations.
This framework further creates a foundation for continuous performance monitoring of the algorithms on future releases and can be easily ported to also test other algorithms (also outside mobgap) to be compared in the same way.
The developed benchmark challenges can also be expanded to further datasets and we actively encourage people to use the structure and tooling we developed as the basis for future benchmarking papers for Lower Back IMU systems.

## Revalidation Results

For all algorithms, we could confirm that the new implementation works at least as well as the old Matlab versions.
For some algorithms, the new algorithms provide a consistent and relevant performance improvement.
We highly encourage you to update your performance expectations based on these new results (see Revalidation section in the documentation).

Compared to the past validation of these algorithms as part of the original Mobilise-D project, the validation performed now is fully reproducible and will be updated when new versions of algorithms or new algorithms are implemented.

## Further changes

With the recent 0.11 and now with the 1.0 release we added substantial tooling to compare and score gait parameters between multiple systems.
These might be helpful far beyond our revalidation study.

We further updated one of the core GSD algorithms tuning it more towards recall (instead of precision), as our initial revalidation results showed that this provides better overall performance when evaluating the full pipeline.

For more details, see the Changelog of the past releases.

As always, we highly encourage you to update to the newest version, as we do not have the bandwidth to support multiple releases of this library.
This means only the latest version will receive bugfixes.

With that... We are looking forward to your feedback and hope that mobgap, and its now fully validated algorithms will be helpful for your scientific work!