By: Oz Moskovich, AI and Data Science Lead, XACT Robotics.
Nearly every sector of healthcare is exploring applications for artificial intelligence, but there are some fields of medicine that present more opportunities for AI disruption than others. As the lead for a data science team in medical robotics, I am keen to find areas of need, and no medical specialty presents a clearer need for AI than interventional radiology.
The challenges facing interventional radiology today include:
- Shortage of specialists: Only about 10 percent of radiologists receive subspecialty training in interventional radiology.
- Cost: The specialist shortage contributes to added costs for patients. Rural patients, in particular, often travel to find the closest interventional radiologist – incurring costs for travel and lodging.
- Timely diagnosis: A recent Sinai study found earlier diagnosis led to a substantial decline in lung cancer deaths.
- Tumor properties: When diagnosing a potential tumor, the size, location and tissue compliance can all lead to delayed diagnosis and treatment.
- Procedure inconsistencies: Manual procedural methods at times require multiple insertions to reach the desired target, which can result in longer procedure times, readmissions or complications.
Fortunately, tools available today are already helping to mitigate those challenges and AI is key among them. By coupling AI and machine learning capabilities with robotic and imaging platforms, our healthcare system can expand access to quality care. That involves improving the speed, efficiency and availability of procedures such as biopsies and ablations, resulting in more positive outcomes and satisfied patients.
Opportunity in robotics
Robotic systems have proliferated across medicine, but the demand for complex and accurate image-guided planning and monitoring in procedures such as biopsies or ablations make robotics an ideal fit for interventional radiology. With accurate, robotic-powered insertion and steering, physicians can diagnose and treat potentially life-threatening diseases earlier – when tumors are smaller and more susceptible to treatment. Robotic technology also provides an avenue to further incorporate AI and machine learning into interventional radiology.
With clinical workflows increasingly incorporating AI-powered technologies in multiple domains, it is just a matter of time for similar adoption of robotic systems. When combined with machine learning, robotic systems can leverage vast amounts of past procedure data to help physicians make highly informed decisions. By sharing that data globally and supplying the means to analyze it, machine learning is becoming a uniting force that gives rise to a more sophisticated level of care grounded in a broader set of experiences. From finding cases with similar characteristics to highlighting risks and anomalies to real-time recommendations, even the most experienced physicians will benefit from access to this set of capabilities. Additionally, pairing AI and imaging produces new capabilities, such as image enhancement, image fusion, tissue segmentation and 3D renderings. Each of those gives the physician the clearest picture of their targets, allows for procedure planning in advance and can contribute to a more precise procedure and optimizes outcomes.
Addressing shortages and inefficiencies
AI-powered robotic platforms have the ability to make procedures more predictable – reducing the risk of a readmission and completing procedures in a consistent amount of time. Part of that predictability is ensuring an optimal outcome with a single procedure and avoiding the need to readmit a patient for a second procedure. Medicare spends about $30 billion annually on hospital readmissions and more than half of that expense goes toward avoidable readmissions. By planning procedures and leveraging big-data, machine learning and AI through robotic platforms, our physicians will execute procedures accurately and efficiently and will reduce wasteful spending on avoidable procedures.
AI also has an opportunity to help solve for specialist shortages. As intuitive devices become more common across healthcare provider facilities and procedural knowledge becomes more accessible, physician extenders – i.e. physician assistants and nurse practitioners – will perform more procedures. By empowering more clinicians with the tools to perform interventional procedures, we can relieve a strained physician population and spread out the clinical burden more equitably.
Applications for AI in medicine remain years away from ubiquity, but ultimately, there is tremendous opportunity for AI to enhance physician capability in interventional radiology – it will never replace them, but rather, will serve as a magnificent new toolbox. By continuing to advance the work that is already in progress across robotics, AI and machine learning development teams, we will introduce cutting-edge technology to interventional radiology. It has the potential to help to solve for a physician shortage and achieve positive outcomes more efficiently and quickly for a larger population of patients.