Synthetic Patients: Can Artificial Data Help Medical Research?

Synthetic Patients: Can Artificial Data Help Medical Research?

By Mansi Goel (Fellow Threws)

Imagine opening a patient’s medical record. It contains their age, medical conditions, medications, laboratory results, and history of hospital visits. Everything looks like a typical medical record, except for one detail: the patient doesn’t actually exist.

This is the idea behind synthetic health data, an emerging area of medical research where artificial patient records are generated to resemble real-world healthcare data.

How Do We Create Patients Who Don’t Exist?

Electronic health records (EHRs) contain valuable information for medical research. However, accessing these records can be difficult because they contain sensitive personal information and are subject to strict privacy regulations.

Synthetic data offers a possible alternative. Instead of sharing actual patient records, researchers can generate artificial datasets that reproduce some of the characteristics and relationships found in healthcare data.

There are different ways to do this. Some methods use statistical models or machine learning to learn patterns from existing datasets and generate new records. Others use medical knowledge, publicly available statistics, and computer simulations to create entirely fictional patient histories.

One example is Synthea, an open-source tool developed by researchers at MITRE and collaborating institutions. It simulates patients’ lives, including medical conditions, treatments, and healthcare encounters, without relying on actual individual patient records. The approach was described in the Journal of the American Medical Informatics Association in 2018 [1].

Why Is This Useful?

Consider a researcher developing software to organize patients’ medical histories. Testing the software requires realistic records containing diagnoses, prescriptions, and laboratory results. But using actual patient information for this purpose may be unnecessary.

Synthetic datasets can provide a practical testing environment without exposing real patient records.

They can also help students learn to work with medical data, allow researchers to experiment with analytical methods, and support early-stage development of healthcare technologies.

In certain research settings, synthetic data may even help supplement limited datasets. However, its usefulness depends heavily on how accurately the generated information represents real clinical patterns [2].

Are Synthetic Patients Really Like Real Patients?

Not entirely, and this is where things become complicated.

A synthetic dataset may reproduce common patterns in a population while missing important details. For example, it might accurately represent how frequently diabetes occurs but fail to capture complex relationships between medications, laboratory measurements, and disease progression.

There is also the question of privacy. Artificial data is not automatically anonymous or risk-free. When synthetic data is generated from real patient records, poorly designed methods may unintentionally reveal information about individuals.

A 2025 review published in npj Digital Medicine examined 73 studies of synthetic medical data. Among 67 studies claiming privacy-preserving applications, only 31 actually evaluated privacy risks [3].

These findings highlight why synthetic data must be evaluated for both its usefulness and its ability to protect patient information.

A New Tool, Not a Replacement

Synthetic patients cannot replace real clinical observations, and findings based on artificial datasets still need appropriate validation using real-world data.

However, they create opportunities for researchers, students, and developers who might otherwise have limited access to medical datasets.

The real value of synthetic data may not be in creating artificial patients that perfectly resemble real ones, but in making it easier to explore new research ideas without unnecessarily exposing sensitive patient information.

References

1. Walonoski J, et al. (2018). Synthea: An approach, method, and software mechanism for generating synthetic patients and the synthetic electronic health care record. Journal of the American Medical Informatics Association, 25(3), 230–238.
2. Giuffrè M, Shung DL. (2023). Harnessing the power of synthetic data in healthcare: innovation, application, and privacy. npj Digital Medicine, 6, 186.
3. Kaabachi B, et al. (2025). A scoping review of privacy and utility metrics in medical synthetic data. npj Digital Medicine, 8, 60.