AWS re:Invent 2025 - Advancing Patient and Business Operations Insights with Gen AI (IND212)

Advancing Patient and Business Operations Insights with Gen AI

Overview

• Presentation by Shane O'Connor and Kendra from Huron Consulting Group, a global consulting firm focused on healthcare and education
• Discussed opportunities for leveraging generative AI to enhance patient experience and business operations analytics in healthcare

Patient Experience Analytics

• Huron's "Rounding" tool uses a structured, evidence-based approach to improve patient experience through meaningful interactions between staff, patients, and families
• Challenges with traditional patient experience surveys:
  • 3-4 month delay in receiving HCAPS survey data, limiting ability to intervene early
  • Need to move beyond basic sentiment analysis to uncover actionable insights
• Generative AI-powered approach:
  • Analyzes survey feedback to identify key drivers of patient experience and HCAPS performance
  • Delivers real-time insights to support staff coaching, service quality improvements, and patient satisfaction
• Architecture:
  • Uses Amazon Bedrock, the Nova LLM model, and Amazon Redshift
  • Sentiment analysis performed on manually entered question/answer text
  • Future plans to automate the rounding process with real-time transcription and analysis

Business Operations Analytics

• Huron's business operations focus on improving client financial performance across revenue cycle, supply chain, pharmacy, workforce, and HR
• Challenges with structured data analysis:
  • Need to integrate disparate data sources to get a holistic view
  • Desire to uncover deeper insights beyond just metric analysis
• Generative AI-powered approach:
  • Integrates multiple data sources (e.g., claims, notes) to curate a comprehensive financial performance history
  • Uses LLMs to analyze unstructured data and highlight areas for improvement in payer relations, staff performance, and denials management
  • Generates staff-level performance scores and overall "effectiveness" ratings
• Architecture:
  • Ingests raw data into S3, processes through Glue and Redshift
  • Leverages LLMs to summarize unstructured data and provide insights
  • Outputs results to visualizations in Amazon QuickSight

Results and Impact

• Sentiment analysis accuracy currently at 90%, with a goal of reaching 99%
• Ability to identify patients at risk of poor experiences and target service recovery interventions
• Potential to improve HCAPS scores and maximize hospital funding
• Scaling to process 10,000 notes per week with 90% accuracy
• Identifying net revenue opportunities and denial trends more effectively

Future Roadmap

• Aggregating sentiment data and emerging themes into BI analytics platforms
• Connecting patient experience and business operations insights to enable more holistic coaching and strategy development
• Exploring opportunities to uncover hidden connections between patient experience, market share, and net revenue

Key Takeaways

• Generative AI enables healthcare organizations to move beyond basic analytics and unlock deeper, more actionable insights from both structured and unstructured data
• Integrating patient experience and business operations analytics can drive significant improvements in care quality, financial performance, and overall organizational effectiveness
• Huron's approach demonstrates the power of combining domain expertise, AWS technologies, and advanced AI/ML capabilities to transform healthcare delivery and operations