Alejandro de la Vega

Improve estimate of compute required

1

Could potentially determine this empirically by trying benchmark models of various lengths on different tasks

We are thinking about the option of expanding Neuroscout to also index a large set of non-naturalistic datasets. However, there is a concern that the ingestion of these datasets could...

Keep track of TR for each Run, versus assuming its the same for all tasks

1

For example, in the `narratives` dataset `milkyway` task, one subject (115) has a TR of 1.52, whereas the rest of the subjects it is 1.5 This causes an issue when...

Currently, all raw nifti volumes are needed upon ingestion to properly record the run duration in the db (which is necessary for accurate reports and design matrices). We can currently...

How to "release" collections of analyses

1

For example, we want to release a collection of "single predictor models" that were run on all Naturalistic datasets on Neuroscout. One possibility is to add `Collections` to NeuroStore, once...

We are working on doing a "data release" of single subject maps. @tsalo asked how he could access these maps and link them to Neuroscout semantics in order to test...

Currently, if I push a non valid `StatsModel` to Neuroscout using pyNS, the fronend will probably crash. Instead, the frontend should validate the `StatsModel` and fail gracefully if validation fails....

Add licenses to feature releases / API

1

It may be a good idea to add a license to the neuroscout results API / release bundles h/t @satra