Hello, I would like to create custom Dataset and my use case is the following:
- I have 75 patients and for each of them I have connectivity graph with signal that changes over time
( so the graph is static, although I have different signals at time t0 and t1)
I suppose they are spatial-temporal graphs and I can use some models from Pytorch Geometric Temporal.
Speaking about the data - how should I load them? For each patient should I save only 1 graph with multiple node features (representing temporal signal) or multiple graphs should be used (separate for each timestep) ?
Disclaimer: I am a newbie in Graph Neural Nets. I do not expect the code but the help with understanding my use case. Any links to similar problem solution’s will be appreciated.