Done (and to do)
- started working on getting SUIT derivatives into BIDS format using pyBIDS, spent some time reading up on pyBIDS as I hadn’t used that library directly. SUIT functionality is still split across matlab and python (segmentation and normalization happen with MATLAB, and only reslicing and mapping from volume to surface happen with Python). Next step is to build a singularity container (something I haven’t done before) that contains both matlab + python packages relevent for running SUIT and output BIDS-like derivatives. ideally, would be great for SUIT to be incorporated into fmriprep pipeline but I think this is beyond the scope right now.
- have been running many predictive models using pydra-ml, both classifier and regression models, predicting disorder diagnosis, demographic info (sex), and general cognitive functioning scores. Most of the models perform similiarly (badly) in terms of accuracy, have been testing out different dimensionality reduction methods (umap, pca etc.) to determine whether this improves model performance (and feature interpretation). Two RAs are starting next week and I will be giving them both projects that focus on building phenotypic models while I turn my attention to neuroimaging. One approach that could be interesting is to relate dimensions of phenotypic assessment to precise connectivity networks (i.e. PFC-Cerebellum), an similar taken by Klein-Fugge et al. (2022)
- one question that I want to address with the HBN data is whether PFC-Cerebellar connectivity can be used to predict clinical symptoms across different developmental stages. Both of these structures have a protracted development, and any dysfunction or dysregulation in this connectivity during critical developmental periods likely contriubtes to symptoms underlying a host of disorders. There are also sex differences in the maturation of PFC and cerebellum, and it would be interesting to test whether connectivity differences emerge at different developmental stages across males and females.
- I’ve also been continuing literature review for the CHOP project and I’m eager to get access to the data so that we can explore relevent questions.
- Onboarded the first of three RAs, here is the onboarding document I have put together.
Questions
- I’m going to send the CHOP team an email today to enquire about next steps (apologies for the delay on that front) - I’ll send it to you first and let me know if there’s anything you’d like to add. I’ll include summary of meeting along with some follow up questions about data, and some specific project goals we can address, most of which is covered here
- perhaps you could send John a line about monthly meetings? I don’t want to bother him with too many emails.