The client is a healthcare software solutions firm with more than 1,500 employees.
The Challenge
In the world of healthcare software, onboarding a new payor to the system requires significant data entry and configuration changes. The client – a major supplier of healthcare software solutions – wanted to offer this additional service to payors to make onboarding easier. However, after configuration, data entry is typically a manual process that doesn’t scale well, causing bottlenecks and slowing the onboarding process. In addition, configuration involves the use of patented Domain Specific Language (DSL) files, and knowledge of HRL (a healthcare-centric DSL) is required to configure and test the system correctly.
The Solution
The Persistent solution used Generative AI to function as transpilers (source-to-source compilers) to generate HRL, a domain-specific language optimized for healthcare business rules and workflows. Transpilers transform the code of a language into another form of the same language. In this case, employing OpenAI GPT 3.5-Turbo APIs on the MS Azure cloud, with custom prompt engineering, enabled a faster and more cost-efficient transpiler approach.
Persistent deployed its PSL Text Analytics accelerator to recognize and standardize concepts, automating the data entry process from an unstructured to a structured format. We also automated the entire onboarding cycle with a maker-checker process for human oversight, using three-step automation: first to generate HRL, then data entry, and finally replicating HRLs for individual configurations and data points.
In addition, the Persistent solution leveraged powerful AI technologies, including ANTLR (a parser generator for reading, processing, executing, or translating structured text), Natural Language Processing (NLP), PyPDF2 pure-Python PDF libraries, as well as Pandas libraries for data manipulation and analysis.
The Outcome
Persistent’s cloud-based GenAI solution delivered 80% faster onboarding of payors, with additional revenue generated from using the DSL platform. This approach has also achieved a 75% savings in total FTE (labor) costs, and no complex knowledge of DSL is currently required.
Technology Used
- Transpilers
- ANTLR
- Natural Language Processing
- MS Azure OpenAI GPT 3.5-Turbo APIs
- Custom Prompt Engineering
- PyPDF2
- Pandas libraries
