We currently support single task supervised training, we should support multiple tasks. Users would input a list of task parquet files (with patient_id, end-point timestamp, and labels) and we should join these dataframes and then integrate them into the pytorch_dataset class here.
We currently support single task supervised training, we should support multiple tasks. Users would input a list of task parquet files (with patient_id, end-point timestamp, and labels) and we should join these dataframes and then integrate them into the pytorch_dataset class here.