Synthetic Data for Healthcare Data at a Hospital Chain Customer
Introduction
Brewdata has a hospital chain customer that collects large volumes of patient data. The data has to be aggregated and submitted to the government for statistical purposes. Due to the Personally Identifiable Information (PII) and privacy rules (GDPR and POPI), the data needs to be cleansed of that information first. This was an onerous process for the hospital chain and with the introduction of the Brewdata suite of products and services has become a more sustainable and disciplined process
A Brewdata customer in the Legal Tech domain had a need to test integrated enterprise systems. This required a very close representation of the Production Data set to ensure that business rules, and exceptions are triggered in the work flow as closely as possible. The use of the Brewdata suite of tools was instrumental in ensuring the Quality Assurance Process of the customer was met to the highest standard of capability.
Use Case
Brewdata’s customer is a hospital chain with several hospitals. Large volumes of patient admissions data is collected on a daily basis. Much of this data has value beyond the immediate treatment of the specific patient. That data is required by the government for statistical purposes and also needed by the hospital chain for planning patient acuity levels, staffing, and capital investments as well as other studies to aid proactive health management of the general population of the regions served by the hospital chain. However patient data has Personally Identifiable Information (PII) and is protected by privacy laws, and cannot be used in its unprocessed form. GDPR and derivative privacy regulations such as POPI, also restricts the use of Production Data in non-production systems. The IT division of the hospital invests in expensive and time consuming processes to transform the data from its original form to that which can be purposed for the planning and reporting activities of the chain. Besides the extensive time for transformation, quality assurance and cost involved, there are additional issues associated with that same transformed data. For instance, the statistical characteristics of that data can potentially change during the transformation process which diminishes the utility of that data. Additionally there is no certification that the transformed data still does not reflect a real person or patient and that is an ongoing risk to the hospital chain. One of the challenges here, is that the PII data can reside in several fields. If for instance a doctor inadvertently mentioned a name or PII in notes, or prescription which is transcribed then it potentially creates exposure of private information. So a rigorous process is needed to ensure that the data is cleansed and transformed.
Approach
The Hospital Chain approached Brewdata for assistance with this use case. As is the standard methodology of Brewdata, the Hospital Chain was first engaged in a FREE discovery workshop to learn more information about the use cases that were specific to the Hospital Chain customer. An initial services engagement helped Brewdata fine tune the Brewdata tools to handle the specific Hospital Chain Customer Use Case and generate the first sets of Synthetic Data. Shortly thereafter with the direct use and leverage of the Brewdata Studio and associated products and services, the Hospital Chain was able to streamline and speed up the process of generating certified Synthetic Data. Brewdata Customer Success teams continued to engage and follow up at regular intervals to understand any unique challenges that were specific to the Hospital Chain data and assisted as necessary to augment the tools or streamline the Synthetic Data generation process.
Benefits
Brewdata’s Customer was able to generate large amounts of test data in record time leveraging the Brewdata product suite and using Production Data as the inspiration and input source for the generation of Synthetic Data. Datasets for different use cases were generated with ease with both the Brewdata Services Engagement and later with the Brewdata Studio product used independently by the customer. No time was spent by the Quality Assurance team in manually generating data to support test cases and instead was spent identifying and documenting quality feedback in the systems integration use cases. And in fact many of unique permutations and combinations of Synthetic data that was generated by Brewdata Studio exposed interventions that were required by the Quality Engineering team to address functionality gaps in the overall enterprise architecture, issues that would not have been uncovered if the Test Data was produced manually.
Opportunities
As the Brewdata customer expands its products and services to the marketplace, new capabilities will be introduced over time and hence the need for additional new Synthetic Data will continue as the customer develops new systems capabilities. Brewdata’s customer has ambitions of venturing into AI capabilities and has the intention of using the Synthetic Data generated for Systems Testing in other areas of value generation including as training data. Hence the value generated by those data sets will be amplified many times over, as well.
Getting in Touch
Brewdata works with customers across many industry domains to provide them with products and services for Synthetic Data generation for many use cases including systems testing. Get in touch with Brewdata at info@brewdata.ai to learn more about Brewdata Products and Services and how we can help you generate Synthetic Data sets to support your specific needs and use cases.