Can AI detect differences in heart tissue related to inherited diseases that are invisible to the human eye? Pernille Heimdal Holm shares results of a study investigating whether AI can help categorise sudden arrhythmic death syndrome (SADS) diagnoses in autopsy cases, as presented at ECP 2026.
Pernille Heimdal Holm is a forensic pathologist at the University of Copenhagen.
The following transcript has been edited for clarity.
My name is Pernille Heimdal Holm, and I'm a postdoc and a medical doctor at the Department of Forensic Medicine, University of Copenhagen, Denmark.
As forensic pathologists, we sometimes investigate sudden and unexpected death in otherwise healthy young people. And if an autopsy finds no cause of death, the heart appears normal, and all additional tests are negative, the case is labeled SADS, sudden arrhythmic death syndrome.
SADS is the most common cause of sudden cardiac death in people under thirty-five. The cases are presumed to have died from a heart rhythm disorder, which can be inherited, but we cannot see it or prove it at the autopsy. This makes it difficult to give answers to the relatives and to know if genetic testing is helpful. Finding even subtle clues in the heart could be extremely valuable and may help us establish a cause of death in the future.
Now, AI-based methods have made it possible to precisely measure specific cardiac changes. So our question in this study was simple: Can AI detect differences in heart tissue related to inherited diseases that are invisible to the human eye?
We included seventy-three SADS cases with a normal autopsy and hearts that all appeared completely normal under the microscope. Six cases had a genetic change in a gene related to a heart muscle or heart rhythm disease. We used an AI-based computer algorithm to identify the individual fat cells and also measure the different types of tissue. This was done in four standard heart samples from each case.
We found that people with SADS who carried a genetic change had a different tissue pattern in their hearts. They had less connective tissue in one area and higher density of fat cells in another. Even though we only had six cases with a genetic change, we could still detect clear differences between these groups.
Interestingly, most of the genetic changes were linked to inherited heart and muscle diseases, and these are conditions that usually cause well-known visible changes in the heart. But in our case, the heart appeared normal, and this suggests that AI-based tissue analysis may be able to detect very early changes that are not yet visible to the human eye. But larger studies are needed to confirm these findings.
Overall, our study shows that AI can help us find hidden patterns in the heart that would otherwise appear normal. Hopefully, in the future, we can then give a diagnosis to what is now deemed an unexplained cause of death.
Newsletters
Receive the latest pathologist news, personalities, education, and career development – weekly to your inbox.
