trial_simulation.rst

Trial Patient Records Simulation

Table of Contents

• Tabular Patient: Index
• Sequential Patient: Index
• Tabular Patient: Example
• Sequential Patient: Example

Synthetic patient records generation is a way around privacy issues when sharing clinical trial records or healthcare data. Specifically, we want to train a generative model based on the real records, as p(h^{syn}|h_1,h_2,\dots,h_n; \theta), where h^{syn} is the synthetic records, h_1,h_2,\dots,h_n are the real records, and \theta are the parameters of the model. When the patient data is a sequence, we can apply the generative model to the conditional generation. Given the previous visits v_{1:t-1}, we can generate the next visit record v_t as v_t \sim p(v_t|v_{1:t-1}; \theta).

Depending ono the input patient data format: tabular or sequence, we have the following two subtasks: trial_simulation.tabular and trial_simulation.sequence.

Tabular Patient: Index

Here is the list of colab examples on each model for this task.

• Model 1: GaussianCopula
• Model 2: CopulaGAN
• Model 3: CTGAN
• Model 4: TVAE
• Model 5: MedGAN

Sequential Patient: Index

Here is the list of colab examples on each model for this task.

• Model 1: RNNGAN
• Model 2: EVA
• Model 3: SynTEG
• Model 4: PromptEHR
• Model 5: KNNSampler
• Model 6: TWIN

Tabular Patient: Example

Here, we highlight the usage of trial_simulation.tabular.CTGAN model for this task.

Sequential Patient: Example

Here, we highlight the usage of trial_simulation.sequence.TWIN model for this task.