Lessons from Evaluating, Implementing, and Adopting Ambient AI

Overview

While ambient AI is generating significant excitement and adoption across healthcare, many organizations and Health Tech Nerds members still have questions about how we go from interest in AI to actual impact. 

This learning event brought together three crucial perspectives - a clinical champion, AI technology provider, and executive buyer - to share real-world insights about the evaluation, implementation, and adoption of an ambient AI assistant at a midwest health system. The conversation featured: 

• Dr. James Blum, Chief Health Information Officer at University of Iowa Health Care

• Dr. Jason Misurac, Pediatric Nephrologist at University of Iowa Health Care

• Alex LeBrun, Cofounder and CEO at Nabla

The discussion provided practical insights and addressed a crucial need as healthcare organizations look to combat clinician burnout while navigating the complexities of AI implementation in healthcare settings. This was not a sponsored event. 

• 00:00 Introduction & Objectives

• 01:40 Understanding the Problem: Clinician Burnout

• 05:31 Evaluating Ambient AI Solutions

• 17:06 Clinician Training and Rollout

• 20:56 Ongoing Measurement and Success

• 31:05 Technical Implementation and EHR Integration

• 39:45 Key Success Factors and Conclusion

Summary & Key Takeaways

Focus on clinician wellbeing vs. productivity

At UIHC, a significant driver for exploring ambient AI solutions was the pervasive issue of physician burnout. Ambient AI offered a potential solution to reduce administrative burden, allowing providers to focus on patient interactions and rediscover the "joy of the practice of medicine," as Dr. Blum emphasized.

• The ambient AI assistant was provided to anyone who wanted to use it, and there were no productivity requirements or mandates attached to usage. 

• They saw burnout reduction from 69% to 43% in their initial pilot group. Burnout was measured using the Stanford Professional Fulfillment Index. See the pilot study results.

• The investment was viewed as improving provider satisfaction and retention. 

Privacy and partnership as top priorities in evaluation

Data privacy and strategic partnership emerged as crucial factors in the evaluation process. UIHC prioritized protecting patient information and sought solutions aligned with their privacy standards. Dr. Blum highlighted UIHC’s commitment to patient data privacy, noting: "We really value the privacy of our patients and we value their data exceptionally." 

• UIHC was concerned about vendors storing full voice recordings, which they considered PHI, even if redacted. They needed a partner who would eliminate data quickly after encounters and rejected many existing AI assistant vendors, whose contracts allow for infinite data storage or model training with patient data. 

• In the evaluation process, UIHC also wanted a true strategic partner to support training and system-wide rollout, and help them move into other areas outside of ambulatory care. They also wanted technology that could help all clinicians, including nurses and allied health professionals, deliver better care without being "tethered to the computer."

• Alex from Nabla highlighted five other key metrics commonly used in the evaluation process: 

  1. Clinical note writing accuracy 

  2. Time saved

  3. Impact on stress level 

  4. Impact on patients 

  5. NPS score 

User-friendly implementation and adoption strategy

Successful implementation hinges on seamless integration with existing workflows and minimal training requirements. User-friendly interfaces and intuitive design contribute to faster adoption rates and positive clinician experiences. Dr. Misurac, shared that their training plan for Nabla was streamlined due to the intuitive nature of the tool: "We really felt like this tool...was extremely intuitive for our users."

• Implementation time was very short. EHR integration was completed in under 10 days with minimal IT resources. Alex mentioned they make EHR integration “as plug-and-play as possible,” and it is “much more straightforward than it was a year ago.” 

• The broader resource commitment for implementation was minimal. UIHC only had to allocate a few days from one engineer to get everything set up and running.  

• UIHC achieved 50% adoption within 8 weeks of launch. Their approach included:

  • Hosting (2) 1-hour live all-staff training sessions, added directly to calendars vs. an opt-in approach and recorded for later distribution

  • Creating and distributing a user guide/manual, quick-start guide, and tip sheets

  • Partnering with PR team and leveraging existing internal communications channels to drive awareness

• Across their clients, Nabla sees adoption up to 80% following initial rollout. There are multiple access methods for the tool including desktop, mobile, and directly within the EHR. 

The power of peer-to-peer influence and widespread applicability

Organic adoption driven by positive word-of-mouth among clinicians plays a significant role in successful implementation generally and across specialties. Alex LeBrun, CEO and co-founder of Nabla, observes: "What works the best is peer-to-peer influence...doctors just listen to doctors.” 

• It also helped that when clinicians got this tool in their hands, they immediately saw the value and there was enthusiasm. The daily time savings is estimated at 1-2 hours per day. 

• Nabla supports 52 specialties. The top two specialties include family physicians and mental health providers. At UIHC, beyond physicians, they’re seeing adoption from dieticians, physical therapists, social workers, and nurses. The tool could also be customized for different specialties’ workflows. 

• As an academic medical center, UIHC had to consider how to balance training medical students and residents while using AI. Dr. Misurac highlighted that there is “a balance between teaching them to use a new technology like ambient AI and not interrupting the learning that comes from mentally processing a history and writing an assessment and plan.” Ultimately, the training programs were allowed to decide how to incorporate the tool into education. 

Key success factors include wellbeing as a value proposition and strong partnership

University of Iowa Health Care's successful implementation of ambient AI technology can be attributed to several key strategic decisions and organizational approaches. Rather than following traditional technology rollouts that focus on productivity metrics and mandatory adoption, they prioritized clinician wellbeing and autonomy, supported by strong organizational leadership and technical execution. Their success factors included:

• Partnership

  • Strong partnership between Nabla and UIHC for implementation, training, and broader adoption

  • Providers saw value in the tool immediately 

• Provider-Centric Approach

  • Framing the investment in ambient AI as a commitment to clinician wellbeing over productivity 

  • Selecting a simple, easy to use tool that integrated into existing workflows and did not require significant training

  • System-wide access rather than limited rollout

  • Focus on provider choice and autonomy instead of system-wide mandates

  • Bottom-up approach where peer recommendation drove uptake 

• Implementation Strategy 

  • Integration ease and a commitment from IT for making this happen 

  • A simple training approach due to the intuitive nature of the tool

  • Customization for different specialties’ workflows

  • Regular monitoring of usage patterns helped identify areas needing additional support