June 24, 2024

HOMERuN Collaborative: Artificial Intelligence Tools for Hospitalist Work

The Hospital Medicine Reengineering Network (HOMERuN) is a rapidly growing collaborative made up of more than 50 Hospital Medicine groups from academic and non-academic hospitals across the United States.

Organizers and Facilitators: Khoosh Dayton, Marisha Burden, Kendall Rogers, Matt Sakumoto, Catherine Callister, Michelle Knees, Kirsten Kangelaris, Angela Alday, Angela Keniston

Background: In recent years, a gradual emergence of artificial intelligence (AI) tools for medical professionals has occurred, notably with the introduction of ChatGPT. Some of these tools include ambient listening devices capable of transcribing patient-clinician encounters, natural language processing tools to help extract information from medical records, clinical decision support programs, and tools that may even facilitate diagnostics.

The current landscape regarding AI utilization and healthcare providers' experiences, in particular among hospitalists, remains largely unexplored. In this project, we aim to explore hospitalists' use of AI tools for work, understand challenges faced in utilizing AI tools, and understand future opportunities to leverage AI to improve patient care and provider satisfaction.

Experience With AI Tools in Work

Clinical Applications

  • Developing clinical deterioration models/predictive modeling, particularly with sepsis, falls, pressure ulcers, readmission risk, discharge date
  • Ambient listening to generate H&Ps
  • AI nudges to improve clinical documentation improvement (CDI) coding
  • Chart summarization

Research Applications

  • Literature review
  • Summarizing qualitative data

Educational Applications

  • Point of Care Ultrasound (POCUS) utilizing Butterfly devices
  • Lecture creation

Administrative and Operational Applications

  • Summarizing meetings
  • Email drafting
  • Using AI tools like ChatGPT to search divisional policies and procedures

"I think if we don't embrace this, you're going to be left in the stone age."

Challenges Faced in Utilizing AI in Work

Privacy and Compliance Concerns

  • Ensuring Health insurance Portability and Accountability Act (HIPAA) compliance
  • Intellectual property issues, such as OpenAI possessing instructors' content

Trust and Reliability

  • Concerns regarding hallucinations and bias
  • Concerns about the accuracy and reliability of AI predictions and outputs
  • Need for constant revalidation

Awareness and Knowledge

  • Lack of awareness and understanding of AI and its applications
  • AI tools requiring extensive editing, making them net neutral in value
  • Resistance to change and acceptance by patients and healthcare professionals


  • Uncertainty about financial support and who will cover the costs

"At this stage, AI doesn't seem to be quite reliable, at least not reliable enough to replace clinician judgment."

Leveraging IT for Work in the Future

Clinical Applications

  • Decreasing documentation burden for notes and discharge summaries to allow for more direct patient interactions
  • Mitigating diagnostic bias and heuristics in medicine.
  • Conducting regression analysis for clinical predictor rules, providing diagnostic percentages
  • First pass review of systems with patients prior to interview
  • Creating summative statements for sign-outs
  • Summarizing paper charts for outside hospital transfers
  • Providing recommendations for care pathways
  • Generating summaries and differentials, recommending diagnostic workups or treatment plans
  • Following up on test results after hospital discharge
  • Managing low complexity patients to help with provider capacity issues
  • Assisting with scheduling follow-ups
  • Using AI overlays on analytics tools to ask questions about population health

Research Uses

  • Pulling data for quality improvement projects, significantly reducing manual review time

Educational Uses

  • Creating individual learning plans in medical education

"I would like to see AI fill in the natural human gaps, serve as the checklist for perhaps those blinders we have for diagnoses we're less familiar with, less aware of, that might pertain to patients on a clinical decision-making end. I would like to see AI minimize to the greatest extent possible administrative burden."

Key Takeaways

  1. AI is being used in clinical, educational, research, and administrative realms.
  2. Major barriers to AI use include privacy, reliability, cost, and knowledge of use.
  3. The future of AI in hospital medicine is limitless. Providers are hopeful that AI will decrease unnecessary or nonfulfilling burdens and increase time on direct patient contact.

Our next meeting will be on July 12, 2024.

Image Attributions: Vector images from vecteezy.com
Check out the HOMERuN website for more information.
If you would like to join the HOMERuN Collaborative calls, please reach out to Tiffany.Lee@ucsf.edu.