Data Analysis: Advocate Health

The Arizona State University Artificial Intelligence Cloud Innovation Center (ASU AI CIC), powered by Amazon Web Services (AWS), collaborated with Advocate Health, the 3rd largest nonprofit health system headquartered in North Carolina with a combined footprint across six states, to develop an AI-driven solution. The aim was to streamline the analysis of extensive employee feedback data, enabling Advocate Health to gain actionable insights that enhance employee engagement and improve overall patient care. This collaboration is part of Advocate Health's commitment to leveraging cutting-edge technology to provide exceptional healthcare services.

Advocate Health faced a significant challenge managing and deriving insights from a large volume of employee feedback collected through surveys. The existing process of manually analyzing this data was time-consuming and often delayed necessary actions. The lack of a unified system to filter and interpret feedback across various dimensions, such as region, department, and tenure, limited the ability to quickly highlight effective strategies, illuminate emerging issues, and acknowledge trends. Advocate Health needed a scalable, automated solution that could provide timely insights to improve employee satisfaction and organizational outcomes.

The ASU AI CIC team worked closely with Advocate Health to create a proof-of-concept (POC) that utilized AI to streamline data analysis and extract actionable insights. This solution focuses on integrating various survey data sources into a unified database, making data access and analysis significantly easier. At the core of this POC is an intuitive chatbot interface that allows employees to engage with the data through natural language queries, simplifying the process of exploring survey results by applying filters such as region, department, and employee level. The system is powered by Amazon Bedrock’s Claude 3.5 Sonnet model, which expertly validates user queries and delivers insights using cutting-edge natural language understanding. Moreover, it automates the generation of actionable recommendations and summaries, providing Advocate Health’s HR and leadership teams with the insights needed to make informed decisions, enhance employee engagement, and drive overall satisfaction. The team clustered the data using the machine learning (HDBSCAN) algorithm through AWS SageMaker processing jobs and Utilized AWS Athena to filter and query the data effectively using AWS Step Functions as orchestration for end-to-end flow. This streamlined approach enables faster, data-driven decision-making, turning complex data into a strategic asset.

The POC developed for Advocate Health demonstrates an efficient way for large organizations to harness data for better decision-making. By automating the analysis of employee feedback, the solution enables Advocate Health to act on emerging trends and address employee needs in a timely manner, ultimately fostering a more positive work environment. This, in turn, contributes to better patient care, as an engaged workforce is optimally equipped to deliver high-quality service.

The integration of AWS services ensures that the solution is scalable and secure, allowing Advocate Health to handle large datasets without sacrificing performance. This collaboration highlights how AI can be leveraged to improve internal processes, creating a model for other healthcare organizations to follow.

"Working with the Cloud Innovation Center was enriching and value adding. The scalable, automated solution they developed allowed us to learn how to effectively apply large language models to manage and derive insights from vast volumes of feedback, while maintaining accuracy, enhancing both satisfaction and organizational outcomes. Their expertise and dedication were evident at every step of the process."  

 Aniket Navalkar, Vice President, Advocate Health

The solution developed for Advocate Health has significant potential for application across other industries. Any organization that relies on large volumes of feedback, such as educational institutions, corporate enterprises, or other healthcare providers, can benefit from a similar AI-driven analysis tool. The Q&A interface can be tailored to specific needs, allowing organizations to gain actionable insights from their data, streamline decision-making, and improve overall engagement.

This solution is especially valuable in sectors where understanding employee sentiment is crucial for maintaining a positive culture and optimizing operational efficiency. The adaptability of the AI model makes it a versatile tool for a wide range of use cases.