How synthetic data is overcoming privacy challenges in healthcare research

The healthcare industry is saturated in data, but accessing the right data, at the right time, remains a persistent challenge.

Stringent, but necessary, privacy regulations around patients’ data often make it difficult for healthcare researchers to access the depth and breadth of data they need to drive new discoveries that improve patient care.

In some cases, strong privacy regulations may “undermine” promising research initiatives in healthcare, according to an article in the Journal of Medical Ethics. For example, projects such as establishing “learning health systems” that require significant amounts of data from both inside and outside of a health system may encounter difficulty in launching due to data privacy restrictions, the article states.

To overcome these limitations, many leading health systems and researchers are embracing synthetic data, which is novel data created to reflect real data without containing identifiable patient information. The statistical properties and intervariable relationships in a synthetic dataset directly reflect the properties of the source data.

As a result, synthetic data has the same utility and can be analysed in the same way as the original, real-world dataset — all the while preserving patient privacy.

Limited access to healthcare data

Accessing high-quality health and healthcare-related data is often challenging due to factors such as cost, patient privacy concerns, and legal or intellectual property restrictions, according to the Office of the National Coordinator for Health Information Technology. To safeguard patient confidentiality, researchers and developers typically rely on anonymised datasets to explore theories, train data models, test algorithms, or build prototypes.

However, anonymised data still carries a significant risk of re-identification — particularly in cases involving rare conditions — which has proven difficult to eliminate entirely. Additionally, interoperability challenges often hinder the integration of data from multiple sources, limiting the ability to thoroughly test analytical models or support the development of software applications.

Synthetic data offers researchers a means of surmounting these obstacles by providing solutions to three key challenges in healthcare data management, according to an article in PLOS Digital Health. First, it enhances privacy protection and safeguards the confidentiality of individual records. By producing artificial information, synthetic datasets reduce the risk of re-identification. 

Second, it facilitates broader and faster access to health data for researchers and other stakeholders. Due to enhanced privacy protections, synthesised datasets can be shared more freely and efficiently. Third, synthetic data helps address the shortage of realistic datasets needed for software development and testing. It offers a cost-effective alternative for developers, enabling more accurate and relevant testing of applications.

The growing availability of public synthetic health datasets and commercially available synthetic data generators reflects the rising demand for accessible, high-quality data. These tools significantly expand opportunities for researchers, data entrepreneurs, and health IT innovators by providing realistic datasets that preserve statistical integrity while maintaining privacy protections, according to the article.

The value of synthetic data

When healthcare organisations can access, explore, and analyse healthcare data without obstacles, delays, or worry, the potential benefits are immense. Following are four key benefits of synthetic data in healthcare: 

1. More efficient processes and resource utilisation: Synthetic data allows users, including non-data analysts, to access and analyse performance trends, identify care gaps, and reduce costs by proactively addressing care delivery risks and inefficiencies.
2. Increased collaboration between stakeholders: By removing regulatory and privacy-related hurdles of data sharing, synthetic data promotes secure and compliant collaboration. Internal and external teams can freely explore and exchange insights, which reduces bias, improves research validity, and accelerates knowledge transfer.
3. Faster research: Researchers can bypass lengthy approval processes and access synthetic datasets immediately. This accelerates the research cycle from weeks and months to hours and days, allowing for rapid hypothesis testing and response, as demonstrated during the COVID-19 pandemic when quick identification of high-risk populations informed public health interventions.
4. Enabling AI development and LLM validation: Synthetic data provides the volume, diversity, and complexity of real-world scenarios needed to build, train, and validate artificial intelligence tools, including large language models (LLMs). It allows teams to generate representative datasets without compromising privacy, supporting rapid iteration and scalable innovation in AI-driven healthcare solutions.

Synthetic data in brain cancer research at McGill University

A significant challenge in clinical neuro-oncology research is the limited availability of data that pertains to rapid-onset conditions with relatively poor prognoses. In search of a solution to alleviate issues associated with a lack of data, researchers at McGill University in Canada assessed synthetic data.

Specifically, they conducted a study that aimed to evaluate the reliability and validity of synthetic data in the context of neuro-oncology research, comparing findings from two published studies with results from synthetic datasets. Researchers created and assessed synthetic datasets for inter-variability and compared them against the original study results.

Researchers discovered that findings from synthetic data consistently matched outcomes from both original articles, and demographic trends and survival outcomes showed significant similarity with synthetic datasets. They concluded that integrating synthetic data into clinical research offers excellent potential for providing accurate predictive insights without compromising patient privacy – particularly in neuro-oncology, given the data challenges associated with the field.

Synthetic data is emerging as a powerful solution to the ongoing challenge of balancing patient privacy with the need for accessible, high-quality healthcare data. By maintaining statistical fidelity while protecting identifiable patient information, synthetic data accelerates research, enhances collaboration, and streamlines care delivery.

Ultimately, synthetic data significantly expands access to a wider range of stakeholders, including teams across healthcare, academia, life sciences, and pharmaceutical research. This broader access helps ignite innovation by enabling more voices, expertise, and questions to shape discovery. As demonstrated in fields like neuro-oncology, synthetic data holds immense potential to transform how we conduct research and develop solutions by doing so faster, safer, and at scale.