## **A Brief Introduction to the Brain Predictability toolbox (BPt)** ##
**Sage Hahn**, University of Vermont Burlington
**Sunday's demo:** the Colab link is [here][1]
**Relevant work:** The Hahn et al. (2021) BPt paper is [here][2] (open access).
**Softwares/programs requirements:** The BPt can be accessed [here][3]. python3.7+ with jupyter notebook installed (see the [first presentation][4]). The code used in this presentation is [here][5] in Google Colab.
**AOMICS dataset:** [PIOP1][6]
**Modalities:** T1w
**Abstract:**
The Brain Predictability toolbox (BPI) represents a unified framework of machine leaming (ML) tools designed to work with both tabulated data (in particular brain, psychiatric, behavioral, and physiological variables) and neuroimaging specific derived data (e.g., brain volumes and surfaces). This presentation will present a worked example of how one could use BP to perform machine learning based analyses on data from the AOMIC: the Amsterdam Open MRI Collection. This example will focus on using IMRI derived "connectome" input brain features in order to predict in a properly cross-validated manner, a range of different individual phenotypes.
[1]: https://colab.research.google.com/drive/1vFGw8HtpDeLCbDmYUiLKwa5JQwsiiUnF?usp=sharing
[2]: https://academic.oup.com/bioinformatics/article/37/11/1637/5995310
[3]: https://github.com/sahahn/BPt
[4]: https://osf.io/teqxb/
[5]: https://colab.research.google.com/drive/1vFGw8HtpDeLCbDmYUiLKwa5JQwsiiUnF?usp=sharing
[6]: https://openneuro.org/datasets/ds002785/versions/2.0.0