Synthetic data as the name suggests is artificial – you may associate the term with more familiar products, for example: materials or food items but the development and use of synthetic data is on the rise.
A search of the internet will give you lots of definition along the same theme “any production data applicable to a given situation that are not obtained by direct measurement" to; data that is artificially created rather than being generated by actual events: either way it is really important in the NHS as we want to keep your data safe and only share it when appropriate to do so.
There are several strategies for building synthetic data: Number Distribution; developed by observation of real statistical distribution and reproducing fake date or generative models, Agent Based: achieved by observing behaviour and producing random data highlighting the effects of the interactions of the agents and finally and Deep Learning models; which are models using variation auto encoders and generative adversarial networks (GAN), these techniques can improve data quality through ingesting large repeated data sets. The development of synthetic data comes with its own unique challenges. It can be time consuming to create and the final quality of the model will have a direct correlation to the quality of the data used as an input. The other challenges are in relation to bias; this may also have been prevalent in the original source data.
The privacy requirements and how your data can be shared means there if often limited availability to support the development of algorithms that can be used to test new products associated with Artificial Intelligence and Machine Learning and this is a good use case of Synthetic data.
The development of synthetic data sets could range from appointment activity to test results. We can use the data sets to support our healthcare professionals and partners with the baseline and development of clinical trials and development of healthcare technology; an area in which is difficult to share real patient data and still maintain patient confidentiality. The NHS has in recent years been introducing health monitoring apps and providing data to train the technology or software could be synthetic.