13260
Blog

A Humane Approach to AI: A Conversation with Dr. Jean Jose of University of Miami Health System

In the dynamic landscape of healthcare, the integration of AI is transforming health systems, particularly in radiology. At the University of Miami Health System, Jean Jose, DO,  Associate Vice Chair of Radiology, is at the forefront of this revolution. We sat down with Dr. Jose to discuss how the health system is leveraging AI to enhance patient care and streamline workflows, especially in the wake of unprecedented challenges.

The Pandemic Catalyst: Accelerating AI Deployment

“We were faced with a couple of peculiar circumstances during and following the pandemic, which really accelerated our AI deployments,” Dr. Jose shared. The combination of a significant population increase in South Florida, a surge in telehealth visits and the rapid expansion of the health system led to an explosion of imaging studies.”

“The increase in volume required us to really work on our efficiency and workflows,” he explained. “We felt that AI could significantly help us, specifically in flagging incidental findings that were critical.”

With radiology boasting the highest number of FDA-cleared AI algorithms, it was a natural starting point for AI integration in healthcare.

Building the Blueprint: A Collaborative Approach

Deploying AI workflows is still uncharted territory for most health systems. Dr. Jose credited the visionary leadership of his University of Miami Health System colleague Alexander McKinney, MD, Chair and Professor of Clinical Radiology, and the collaborative efforts of a dedicated committee. The team worked to develop internal, institutional solutions.

“We needed to have that common understanding, so that our administrative leaders, people beyond radiology, our colleagues and other specialties really understood what we were talking about, and what resources we would need,” Dr. Jose emphasized.

Focusing on Actionable, Critical Findings

Dr. Jose and his team developed a unique approach to AI-generated workflows, emphasizing institutional governance and tailored solutions. “We strongly feel that the deployment of AI requires institutional governance and the development of workflows that are specific to that institution because no two practices are the same,” he noted.

They devised specific definitions to categorize AI findings:

  • Actionable: Findings requiring intervention.
  • Non-actionable: Findings with no significant impact on patient care.
  • Critical: Actionable findings requiring urgent intervention.
  • Incidental: Unexpected findings from an imaging study.

This led to the creation of five broad categories, each dictating the type of point-of-care deployment.

Point-of-Care Workflows: A Patient-Centric Approach

Dr. Jose walked us through their innovative point-of-care workflows, highlighting the importance of human intervention.

  • Category 1: Actionable, non-incidental, critical findings — e.g., intracranial hemorrhage (ICH). Nurse practitioners hold patients and coordinate with referring clinicians.
  • Category 2: Actionable, incidental, critical findings — e.g., pulmonary embolism (PE). Nurse practitioners monitor and stabilize patients, leading to faster treatment times and decreased mortality.
  • Category 3: Actionable, non-incidental, non-critical findings — e.g., brain aneurysm (BA). Traditional reporting pathways are used.
  • Category 4: Actionable, incidental, non-critical findings — e.g., coronary artery calcification (CAC). Nurse practitioners contact patients post-scan to ensure proper follow-up.
  • Category 5: Non-FDA cleared algorithms (currently not deployed).

“That added human touch to it is so important to our patients,” Dr. Jose shared, recounting patient testimonials of gratitude for humane and effective care.

Research and Impact: Decreasing Turnaround Time (TAT)

Dr. Jose presented preliminary data from their research, showcasing the significant impact of human intervention in AI-driven workflows.

“What the data is showing us is that we have a significant prioritization impact when we have that human intervention,” he explained.

The research demonstrated a dramatic decrease in report TAT and time-to-treatment when nurse practitioners were involved.

“Humans will respond more attentively to other humans, as opposed to getting bombarded by constant notifications,” Dr. Jose noted

Advice for Healthcare Facilities: Intelligent Deployment

Dr. Jose offered valuable advice for healthcare facilities looking to integrate AI:

  • Engage stakeholders with AI expertise.
  • Develop internal governance and controlled pilots.
  • Tailor algorithms to specific patient populations.
  • Invest in the necessary infrastructure.

“Don’t turn on everything at once,” he cautioned. “And don’t turn on everything at once in every environment.”

Looking Ahead

The University of Miami Health System is setting a new standard for AI integration in radiology, prioritizing the human element of patient care and building out efficient workflows. Dr. Jose’s insights underscore the importance of thoughtful deployment and human-centered technology. As AI continues to evolve, the lessons learned from this pioneering approach will undoubtedly shape the future of healthcare.

Explore the Latest AI Insights, Trends and Research

Marlee Ravid
Marlee Ravid brings over a decade of experience in content marketing, communications and customer engagement to her role as Customer Marketing Manager at Aidoc. She strategically executes innovative programs that amplify the leadership of both Aidoc and its customers, helping to position them as visionaries in healthcare. Having been with Aidoc since its early days, Ravid has worked closely with leadership to build and implement comprehensive marketing strategies, from content development to demand generation and brand awareness. She holds a bachelor’s degree from Georgia State University and a master’s degree from Tel Aviv University.
Marlee Ravid
Manager, Customer Marketing