Alzheimer's classification from fMRI
- Building a pipeline that turns raw fMRI brain scans into functional connectivity graphs, where each node is a brain region and the edges are how correlated their activity is.
- Training Graph Convolutional Networks on those graphs to classify Alzheimer's.
- Comparing Pearson, Spearman and Kendall correlation to see which connectivity measure gives the model the most useful graph.
nilearn · PyTorch Geometric · Python
Predicting student well-being (StudentLife)
- Working with StudentLife, a dataset of phone sensor data (activity, sleep, location, etc.) collected from college students over a term.
- Built a pipeline that cuts the continuous sensor streams into sliding-window sequences.
- Trained LSTMs on those sequences to predict students' well-being and academic outcomes.
PyTorch · LSTMs · Pandas