1. insights
  2. the academy 360
  3. artificial intelligence
  4. how one system dramatically cut pre visit burden with custom ai
Academy 360 | the-academy-360

How One System Dramatically Cut Pre-Visit Burden with Custom AI Summarization Tool

Graphic with a deep blue gradient background displaying the word “Insights” and “360°” alongside The Health Management Academy logo in the bottom left corner.
Related Topics:
Artificial Intelligence
Relevant Roles:
Chief Nursing ExecutivesChief Nursing OfficersChief Transformation Officers Chief Learning Officers Chief Information OfficersChief Operation OfficersChief Physician ExecutivesChief Strategy Officers

To explore AI’s growing role in healthcare, THMA’s AI Catalyst team spotlighted City of Hope’s innovative use of AI in oncology. Through AI-powered summary tools, City of Hope reduced pre-visit summary burden from hours to just minutes. Driving progress via their in-house HopeLLM model, City of Hope customized their solutions to fit into their oncology niche. The resulting AI-generated summaries demonstrate 99% accuracy while dramatically reducing clinician review time, allowing more focus on direct patient care and clinical decision-making.

City of Hope’s use of AI highlights the transition to a new era of patient documentation, reflecting how technology can help power efficient, patient-forward solutions. As AI continues to gain traction in health systems, THMA’s City of Hope case study underscores the methods, metrics, and trends health systems should know amidst the technology boom.

1. Filling in the Gaps with a Unique LLM

  • Streamlining repetitive, time-consuming processes, HopeLLM fills in the vendor gap. In response to an unfruitful search for an AI model to process their clinical information from a wide range of databases, City of Hope developed their own large language model, leveraging their existing HopeBERT, trained on millions of oncological clinical notes. HopeBERT’s integration into the HopeLLM provided a thorough, tailored solution to optimize administrative processes.

  • By refining documentation processes, clinicians can focus on patient care. Designed to enhance efficiency and reduce clinicians' cognitive load, HopeLLM enables clinicians to focus on care rather than record review. It now supports multiple use cases—pre-visit summaries, Q&A, research abstraction, trial matching, and patient journey mapping—with strong clinician feedback and measurable gains in efficiency.

2. What it Does

  • HopeLLM provides clinicians with comprehensive patient information, histories, and summaries. HopeLLM minimizes the risk of missing critical information by analyzing treatment histories with multiple queries and AI agents for fact-checking, relevance, and completeness. It captures progression patterns, plan modifications, recent surgical interventions, and other essential oncological events

  • Organizes patient events, timelines, and conditions help providers find the most relevant information. With options to toggle between detailed narrative events and narrowed patient history, City of Hope’s modernization of documentation synthesizes enormous amounts of data across a variety of sources (EMR, imaging reports, non-digitized clinical notes).

A360 Takeaways for Other Health Systems:

  • Invest in Specialty-Specific AI Foundations. Health systems aiming to adopt AI in areas like oncology need more than off-the-shelf tools. Success depends on building strong data infrastructure, engaging clinician experts, and cultivating AI talent to address complex specialty workflows beyond what primary care requires.

  • Target the Right Documentation Pain Points. To get the most out of their AI infrastructure, health systems should analyze each specialty’s workflow to find the biggest time sinks. In the case of HopeLLM, pre-visit summaries—requiring synthesis of diverse clinical data, offered greater opportunity for impact than note generation alone.

  • Plan for Rigorous Validation and Governance. To ensure safety, accuracy, and adoption, health systems must integrate thorough validation, iterative clinician feedback, and strict governance on privacy, compliance, and safety, accepting that these steps often take longer than building the AI itself.

These results demonstrate that front-end AI solutions can significantly improve clinical workflows while maintaining the high standards of accuracy and reliability essential for effective oncology care.