Hi everyone! This post is to give an update on PIE: The Parkinson’s Insight Engine code repository created and stewarded by @cameronreidhamilton with support from @ehutchins. The original post about PIE can be found here: https://rcop.michaeljfox.org/t/parkinsons-insight-engine-pie/962, and a webinar on its structure and usage is here: https://zenodo.org/records/17049395.
Since then, and based on excellent feedback from the community, we have separated the code into two parts:
- PIE is still the Insight Engine, which automates machine learning on the PPMI data
- PIE-clean, which is focused on data loading, cleaning, and merging
Originally, the PIE-clean functionality was included in PIE as layers of functionality which would automatically run as part of the ML pipeline. By separating PIE-clean into its own library, we hope that its functionality will be clearer and easier to use for those who simply want to explore the PPMI data, without requiring ML functionality. And PIE still automatically uses PIE-clean for data loading and cleaning, so there is no change in usage for those who do want ML capabilities!
In terms of future development, the high-level plans are as follows:
- PIE: creation of a GUI to simplify machine learning
- PIE-clean: expand the documentation and use cases, to make it as easy as possible to get up and running with PPMI data cleaning and preprocessing capabilities
However, we are very open to suggestions and feedback! If you have other ideas or requests, please either comment here, submit a request on GitHub, or speak to Cameron, Elizabeth, or myself. We’d love to hear what else you want from PIE!
