Researchers at Boston’s Massachusetts General Hospital have developed software that provides evidence-based automated support to physicians for diagnosing the cause of ischemic stroke.
However, determining the cause of such strokes—which occur as a result of an obstruction within a blood vessel supplying blood to the brain—is no small feat.
That’s because there are more than 150 different abnormalities that are potential causes—or etiologies—of ischemic stroke, and about half of patients exhibiting symptoms show more than one possible cause, says Hakan Ay, MD, a vascular neurologist and director of stroke research at Mass General’s Martinos Center for Biomedical Imaging.
The software reduces the complexity in determining the cause of a stroke by leveraging classification criteria that are well defined, replicable and based on evidence rather than subjective assessments. The software, called Causative Classification of Stroke (CCS), leverages an algorithm that has the ability to generate categories of etiologies with different clinical, imaging and prognostic characteristics.
Ay says the software was developed internally at Mass General more than 10 years ago, but has gone through subsequent modifications. He describes the software as a “living algorithm” that “can incorporate new information as it emerges.”
For instance, Ay notes that new etiology-specific biomarkers, genetic markers, imaging markers and clinical features can be added into the existing CCS algorithm to further enhance its ability to determine the underlying causes of stroke. Nonetheless, he contends that the software’s basic framework has not changed over time.
“There is no other algorithm—commercial or academic. It is the only software of its kind,” says Ay. “It’s a data collection and storage platform as well.”
In a study of 1,816 patients with ischemic stroke recently published online in JAMA Neurology, Ay and his colleagues demonstrated that the CCS provided better correlations between clinical and imaging stroke features and were better able to discriminate among stroke outcomes than were two conventional, non-automated classification methods. CCS was able to assign etiologies to 20 percent to 40 percent of the patients for which two other systems were unable to determine a cause. In addition, the software was better at determining the likelihood of second stroke within 90 days.
He points out that the two conventional classification systems—Trial of Org 10172 in Acute Stroke Treatment (TOAST) and ASCO (an acronym for atherosclerosis, small-vessel disease, cardiac source and other cause)—are not automated like CCS and depend largely on physicians’ opinions. Those systems have “written rules and their own criteria, but a physician incorporates all the information from tests and other sources,” according to Ay.
Because TOAST and ASCO rely on physicians’ experience and knowledge, their results are highly variable, he adds. In fact, Ay reveals that the disagreement rate among physicians in determining the cause of strokes can be as high as 50 percent, which “creates a lot of confusion.” However, he says previous studies have shown that the use of the CCS algorithm reduced the disagreement rate among physicians from 50 percent to about 20 percent, which Ay calls a more reasonable level of disagreement.
The software has been adopted not only at Mass General but by academic centers nationwide, as well as by international stroke registries. “It is being used by investigators widely, both at MGH and elsewhere,” adds Ay, who contends the software serves as an important research tool by providing investigators with both the ability to examine how stroke etiologies interact with one another and the flexibility to define new etiology subtypes.
Licensed by Mass General, CSS is free for academic use and is available here. The software works on any Internet-connected platform—all that is required to run it is a web browser.
Going forward, Ay says his team is working to make the algorithm more user friendly. In particular, the goal is to have the software automatically populate data entry fields by linking to electronic health records. “It is envisioned as a confidential interface that will automatically get the information from an EHR,” he concludes.
Register or login for access to this item and much more
All Health Data Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access