When we started building a comprehensive normal database (ND) feature for Sift, we answered the questions that you face when trying to summarize biomechanical signals. How do we build a ND? Is there a formula for doing so? How do we determine how many subjects are enough? These questions echo the sentiments shared in a previous HAS-Motion blog post, “The Challenges of Assessing an Individual Against Normative Data“, where our own Amy Coyle touched on some of the uncertainty around what is a good ND. She discussed the challenges of deciding who and what to include, the specific activities subjects should perform, and the variety of demographics that should be represented. These questions are central to the development of NDs because they influence how we can use the database in our analysis both the quality and applicability of the dataset for our comparisons.
So Sift’s ND feature was built with all of these questions in mind. We’ve provided our users with a robust way to summarize data sets and to incorporate these summaries into future analyses. And this feature is flexible enough that you can tailor your databases on the go, picking out those key summary statistics and signals for a comparative analysis without needing to write your own code. So let’s get into the details of normal databases in Sift!
Shaping your Normal Database in Sift
The ND feature builds off of Sift’s CMZ library, which already allows users to gather their data from multiple sessions into one place. From there, you can specify exactly which signals you want to include and exactly which statistics you want to calculate. These calculations happen directly in Sift, eliminating the need to write custom code in another application.
Another critical aspect of building an ND is deciding how much information to store and how much summarizing is necessary. How do we retain enough detail for the data to be useful while also ensuring the database remains manageable and accessible?
Sift’s ND feature gives you the ability to choose the level at which you want to summarize your data:
- Library summary – a comprehensive overview of an entire project across all participants and motion capture sessions
- Workspace summary – a more granular summary for each motion capture session; this can also line up with each participant
- Trial summary – a detailed summaries of individual trials performed by each of your participants
With Sift‘s new ND feature, users can now effortlessly summarize entire libraries using customizable and fully documented normal databases. Thanks to its user-friendly design, you can easily tweak your ND until it meets your exact needs.
Using your Normal Database in Sift
The ND file format is self-documenting, which makes it easy to share within your lab or externally with other collaborators. Having an ND format that describes the included signals, statistics, and summary levels allows you to make meaningful, long-term comparisons and makes your analysis reproducible.
Once your ND is loaded into Sift, this self-documentation allows the application to provide you with clear insights into how the ND was defined. You can plot, explore, and analyze signals from the ND while keeping them distinct from other CMZ files in the library and you can load multiple ND libraries simultaneously. Sift queries the ND signals and metrics of your choice, displaying them with a distinct “N” group type to represent the normal database. Treated similarly to other groups, ND queries are accessible throughout the application for analyses like PCA, SPM, and more.
Sift users will see how the new ND feature seamlessly integrates with their existing analysis workflows, allowing them to leave custom code and spreadsheets behind.
Conclusion
We are incredibly excited to introduce this new feature in Sift, giving you flexibility to incorporate custom normal databases with the rest of your biomechanical analysis.. With Sift‘s new ND feature, you can now shape, query, and share normal databases, making your biomechanical analyses more efficient when working with large datasets. We look forward to seeing how our users build their normal databases and bring them into their analysis workflow.
Interested in trying this out? Let us know how you would customize your ND for your research project! Contact info@has-motion.ca for more information.
