Strata Decision Technology Innovation, Technology & Agility

How is your team integrating AI and ML into the product development process, and what specific improvements have you seen as a result?

We have been using AI and ML models for years at Strata. I have also been lucky to be part of the pilot team at Strata to figure out how to use Generative AI in our day-to-day work. 

In ideation, ChatGPT assists in brainstorming and refining ideas, streamlining the creative process. For user research, we leverage Dovetail AI to quickly transcribe, summarize and analyze insights, allowing us to make faster, data-driven decisions. GitHub Copilot accelerates development by providing real-time code suggestions, cutting down repetitive tasks and expediting the build phase. 

One of our data products, the Comparative Analytics tool, uses advanced ML techniques that normalize and integrate customer internal data with external data, making comparisons of financial data easy and fast. By standardizing data automatically, it removes the manual data submission, so customers can pull up metrics and compare with any peer group they choose. This is now even more important as we are integrating multiple data products that Strata has acquired. We are in the process of coming up with new standards for our next generation of products.

What strategies are you employing to ensure that your systems and processes keep up with the rapid advancements in AI and ML?

We are actively piloting proofs of concept for GenAI tools and LLMs to explore emerging technologies, supported by an AI advisory group that monitors and evaluates new developments. We also work with our governance, regulatory and compliance team to ensure our data is safe throughout the lifecycle of each AI tool and ML model from POC to implementation. We have a very exciting dataset and we take data security and privacy very seriously. 

We also seek partnerships for new technologies and LLMOps platforms, expanding our capabilities as AI evolves. To maintain ML model relevance, we continuously update and monitor our ML models for performance impacts at the product level. 

Additionally, we’ve centralized AI knowledge, best practices and standards across the organization, created extensive documentation and training resources, equipping new users to adopt AI tools effectively. Finally, a cross-functional group ensures AI tools are implemented and utilized properly across departments, maximizing consistency and impact across the organization. This multi-layered approach keeps us agile, secure and aligned with the latest AI advancements.

Can you share some examples of how AI/ML has directly contributed to enhancing your product line or accelerating time-to-market?

One example is how we use ML models to normalize our data. To create a true “apples to apples” comparison between our customer and their peers, our data products utilize discriminative ML models during both implementation and production stages. These models standardize key categories, such as job titles, department names, account details and payor information to ensure consistent data across varied datasets. Healthcare data can be incredibly complex. For example, we found over 700 variations just for the title “Nurse” in our dataset. Using ML to standardize job titles allowed us to accurately calculate the nursing turnover rate and track how this rate shifted during the pandemic. 

During our GenAI tool proof of concept, we closely tracked the time saved using AI tools. A prime example is synthetic data creation for testing, a traditionally labor-intensive process involving manual data adjustments or custom tooling for specific configurations. ChatGPT reduced time spent on this process by 92 percent, generating synthetic data files or skeleton scripts as needed. Engineers highlighted that AI tool automation relieved them from tedious tasks, enabling them to focus on more complex work.