Tutorial (notebook) showing how to use a trained model for inference time

li-lab-mcgill / GAT-ETM

"Modeling electronic health record data using an end-to-end knowledge-graph-informed topic model" paper on Sci Rep (2022)

https://doi.org/10.1038/s41598-022-22956-w

8 stars 0 forks source link

Open gmgeorg opened 1 year ago

gmgeorg commented 1 year ago

After training the model (and storing torch model to disk), it would be very helpful to have a tutorial (notebook) that shows how to

load a trained model
run .predict_topic() on a new EHR record with a set of ICD and ATC codes (e.g., ["A10", "M54"] and ["N02"]) for topic modeling
run .predict_atc() for the ATC recommendation example
run .predict_icd() for an ICD recommendation example
show how you obtained visualizations you do in paper

gmgeorg commented 10 months ago

@zouyuesong any pointers here?