Conditional GAN for better embedding and generation of medical codes
| Title | Conditional GAN for better embedding and generation of medical codes |
|---|---|
| Summary | Synthetic data generation of Electronic Health Records with a focus on medical codes |
| Keywords | GANs, Electronic Health Records, Representation Learning |
| TimeFrame | |
| References | Data: https://mimic.mit.edu/docs/about/
papers: https://dspace.mit.edu/handle/1721.1/128349 https://proceedings.neurips.cc/paper/2019/file/254ed7d2de3b23ab10936522dd547b78-Paper.pdf https://www.sciencedirect.com/science/article/pii/S0957417421000233 |
| Prerequisites | |
| Author | |
| Supervisor | Stefan Byttner, Amira Soliman, Kobra Etminani, Atiye Sadat Hashemi |
| Level | Master |
| Status | Open |
The use of Electronic Health Records (EHR) is increasing in both research and clinical practice to enhance the ability to provide the needed care to patients without posing additional economic burdens. Patient data is considered temporal data that is being measured over time. Due to privacy constraints, EHR data can’t be publicly shared for research in machine learning and artificial intelligence. Synthetic data generation introduces a solution, by generating artificial patient data. This thesis aims to investigate the use of generative adversarial networks (GANs) in better representation and generation of patient encounters with medical diagnoses. Example of research questions: How to represent medical codes as categorical variables? How to enrich the representation of categorical variables?
