All our data are preprocessed into 0.1Hz signals of length of 180 time steps.
The DANP model in src/model/DANP.py
takes in three pytorch tensors as input: train_real.pt
, test_real.pt
, and sim_data.pt
. Each of the three tensors should be of dimension [N, 180, 7]
, where N
is the number of rows, depending on source data. The 7 columns are features in the following order:
[MAP, motor_speed, pump_flow, LVP, heart_rate, tau_lv, contractiity]
Please see the paper for more details on preprocessing.