A common test could now reveal hidden heart trouble
Dr. Venkatesh Murthy, a researcher at the University of Michigan, has developed an artificial intelligence tool to detect a specific type of cardiovascular disease known as microvascular dysfunction. This condition causes chest pain and is especially prevalent among women, according to Murthy. The standard method for diagnosing the condition is through a cardiac PET scan, but these tests are not widely accessible or affordable. In many areas, patients may have to wait three to four months to get scheduled for one. Murthy's new AI model uses data from electrocardiograms, a much more accessible and common diagnostic tool, to detect signs of the disease more quickly. This could significantly reduce the time it takes for people to receive a diagnosis and begin treatment.
In the initial tests of the AI tool, it correctly identified the condition in 75% of cases. Murthy believes that with further development and refinement, this accuracy could rise above 90%. Although the tool is not yet approved by the FDA, it could still help fill the gap where high-quality diagnostic resources are lacking. If it proves to be reliable and effective, it could be a game-changer for patients who are waiting for access to the more advanced testing options.
Why EKGs beat waiting lists
PET scans are not only costly but also difficult to obtain, with only around six such facilities available in the state of Michigan alone. For patients who live in areas without a PET scan center, the options are limited. The long wait times can cause delays in care and result in some patients going undiagnosed for years. Murthy’s AI model offers a new alternative by using data from electrocardiograms, which are much more affordable, widely available, and can be conducted in nearly every medical office or hospital. By analyzing EKG results, the model can predict which patients are most likely to need a PET scan and which might not need one at all, helping to reduce the number of unnecessary tests and saving both time and money.
This strategy is especially important for diseases that are either rare or require complex diagnostic methods. Murthy emphasized that the team's goal is to develop tools that can be used for both common and uncommon conditions, ensuring they are practical and accessible. The AI model, he said, is not just a prototype—it's intended to be deployed in clinics and hospitals around the world wherever EKG equipment is present. That means patients in remote or underserved areas could benefit just as much as those in major medical centers.
What’s next for the AI model
The next phase of the project involves testing the AI model with data from a much broader and more diverse patient population. Murthy’s team is partnering with the Wellcome Leap VISIBLE program to expand the model’s training and validate its effectiveness in different settings. This will be an important step in making sure the AI can work well across a wide range of demographics and health conditions. The ultimate aim is to further boost the model's accuracy and ensure it is reliable in real-world use, regardless of where or on whom it is being applied.
Murthy explained that his team's approach to building the AI model was driven by the need to do more with less. Rather than relying solely on large data sets, the team used a combination of clever design and innovative techniques to create a powerful tool from a more limited set of information. This strategy allows for the development of diagnostic tools that can be used in a variety of situations, especially in areas where data or resources are scarce.
Microvascular dysfunction is a condition that affects the small blood vessels in the heart and other parts of the body. Over time, it can lead to serious health problems, including heart failure and dementia. By detecting the condition early with an accessible and efficient method, Murthy’s AI model could help stop these more severe issues before they become life-threatening. This approach not only improves individual patient outcomes but also reduces the overall burden on the healthcare system by catching problems before they escalate.