In the rapidly evolving landscape of healthcare technology, James Maitland stands as a pivotal figure bridging the gap between sophisticated data analysis and life-saving clinical applications. With a robust background in robotics and IoT, Maitland has dedicated his career to transforming how health systems leverage artificial intelligence to move from reactive treatment to proactive prevention. His work focus centers on “opportunistic screening”—using the vast amount of existing medical data to catch silent killers like heart disease before they manifest as emergencies.
We spoke with Maitland about the integration of AI in cardiovascular care, the economic shifts brought on by new reimbursement codes, and how “agentic AI” is reshaping the administrative and clinical workflows of modern hospitals.
Routine chest CT scans often contain hidden data, such as coronary artery calcium, that frequently goes unaddressed. How does your screening process flag these incidental findings, and what specific clinical triggers ensure a patient actually reaches a cardiologist for follow-up?
The process begins by tapping into the millions of chest CT scans already being performed for non-cardiac reasons, such as pneumonia or trauma. Our AI algorithms, which hold nine FDA clearances, automatically scan these images to detect and quantify coronary artery calcium (CAC) and aortic valve calcium (AVC) without requiring extra radiation or appointment time. Once the AI identifies clinically significant findings—which occurs in roughly 11.7% of routine scans—it triggers a multi-step clinical pathway. First, the finding is surfaced to a physician for adjudication to ensure the AI’s data aligns with the clinical context. Following this, the system uses automated chart reviews to see if the patient is already on a statin or under a cardiologist’s care; if not, an AI agent triggers a referral and coordinates with the care team to schedule a follow-up.
Many high-risk patients, including those in high-stress professions, show no symptoms until a major cardiac event occurs. Can you share a specific instance where AI identified a critical blockage in a major artery before permanent damage happened, and what metrics illustrate the impact of these interventions?
A particularly moving case involved a police officer at the University of Texas Medical Branch who underwent a routine scan that had nothing to do with his heart. Our AI solution detected severe warning signs in his “widow-maker” artery, a blockage that is often fatal and frequently asymptomatic until it is too late. Because the AI flagged this incidental finding, the officer underwent a successful revascularization procedure before any permanent heart muscle damage occurred. Beyond individual stories, the metrics are staggering; in our NOTIFY-1 study at Stanford, using these AI notifications led to a 44-percentage-point increase in statin prescriptions. This demonstrates that we aren’t just finding data—we are fundamentally changing the preventative treatment trajectory for nearly half of the at-risk population identified.
The introduction of national billing code G0680 provides a $15.50 payment for algorithmic analysis of existing scans. How does this reimbursement change the economic viability of preventative cardiology, and what hurdles remain for health systems trying to scale these programs across tens of thousands of annual scans?
The G0680 code is a landmark development because it provides a dedicated reimbursement pathway for the algorithmic analysis of CAC and AVC, effectively “de-risking” the technology for hospital administrators. Previously, health systems had to absorb the incremental costs of these programs entirely, but this $15.50 payment helps recoup the costs of running the analysis at scale. However, the hurdle isn’t just financial; it’s operational. When a system runs 100,000 scans and the AI identifies 9,000 at-risk patients, you cannot simply hand a list of names to a manual care team and expect results. The challenge lies in the “political capital” and change management required to build the infrastructure that can actually handle that sudden influx of thousands of new patients.
Simply surfacing risk data can overwhelm care teams if they must manually review every chart. How do you use automated AI agents to handle triage, prior authorizations, and patient outreach, and what anecdotes show how this reduces the administrative burden on clinical staff?
We address “alert fatigue” by moving beyond simple detection to a platform we call Carebricks, which utilizes “AI agents” to orchestrate complex tasks. These agents don’t just flag a problem; they perform a deep dive into the electronic health record to verify if a patient’s risk factors are already being managed. For example, at UTMB, we have over 20 agents live that handle everything from prior authorization automation to escalating high-risk referrals to the top of the queue. This reduces the manual “paper-chase” for nurses and coordinators, ensuring they only spend time on the most critical cases that truly require human intervention. By automating the outreach and scheduling, we ensure that “access to care” becomes a reality for patients who might otherwise fall through the cracks due to administrative bottlenecks.
Data indicates that notifying clinicians of incidental heart findings can lead to a 44-percentage-point increase in statin prescriptions. Beyond medication, how do these AI-driven insights reshape the long-term management of at-risk populations, and how do you track the resulting improvements in patient health outcomes?
These insights move the needle from reactive “sick care” to true population health management by catching patients “higher up in the funnel.” When we identify calcification early, it changes the conversation from emergency surgery to lifestyle modification, regular monitoring, and preventive pharmacy. We track success by monitoring longitudinal metrics, such as the rate of subsequent major adverse cardiac events (MACE) and the adherence to follow-up appointments scheduled by our agents. The goal is to bend the curve of heart disease mortality across an entire region, using the AI to ensure that no patient with a “ticking time bomb” in their chest is ignored simply because their scan was for a broken rib or a cough. It turns every routine encounter into a comprehensive health screening.
Moving from narrow, single-use tools to broader platforms allows for faster iteration in a clinical setting. How does an “agentic AI” strategy allow you to orchestrate complex workflows like aortic aneurysm follow-up, and what advice do you have for hospitals trying to choose between different AI architectures?
An agentic AI strategy is superior because it focuses on the entire workflow rather than a single point of data; for an aortic aneurysm, the AI doesn’t just measure the size, it tracks growth over time, checks previous scans, and alerts the surgeon if the growth rate hits a specific threshold. My advice to hospitals is to avoid “point solutions”—tools that only do one specific thing—and instead double down on flexible platforms that can be iterated upon quickly. As frontier models like GPT or Gemini become more capable, the bottleneck in healthcare will be how fast you can deploy and adjust your workflows. Look for an architecture that is “iterable and responsive,” allowing you to try new clinical pathways responsibly while focusing on incremental, disciplined improvements to the patient journey.
What is your forecast for AI-powered cardiac screening?
I believe that within the next five years, “opportunistic screening” will become the standard of care for every medical imaging study performed in the United States. We are moving toward a future where it will be considered a clinical oversight not to use AI to look for secondary risks like heart disease or osteoporosis when a scan is already available. As reimbursement pathways for AI expand beyond the current 26 CPT codes, the economic incentives will finally align with public health goals. We will see a massive shift where hospitals function as proactive health hubs, using “frontier intelligence” to ensure that a patient’s visit for a minor issue results in the lifelong management of their most significant underlying risks. This isn’t just about better software; it’s about a fundamental redesign of how we prevent the world’s leading causes of death.
