Table 4 Training (pre-training + joint) and inference times of ClinicalGAN. sim refers to individual patient level simulations.
From: ClinicalGAN: powering patient monitoring in clinical trials with patient digital twins
Attribute | Value |
|---|---|
Training time calculation | |
No. of epochs (pre-training components) | 2500 (each for AE, S and \(T_r\)) |
No. of epochs (joint components) | 10,000 |
Training time (Total) | \(\sim 5.5\) h |
Training time (epochs/s) | \(\sim 0.9\) epochs/s |
Inference time calculation | |
No. of unique patients to be simulated (test) | 250 |
No. of simulations per patient | 1000 |
Inference time (total) | 158 s |
Inference time (ms/simulations) | \(\sim 0.632\) ms/sim |
Inference time (simulations/s) | \(\sim 1585\) sims/s |
