AI-driven stroke tool shows improved outcomes in major clinical trial

• A new study suggests that a stroke clinical decision support system (CDSS), which uses artificial intelligence (AI) assisted imaging, could help to significantly reduce the risk of recurrent vascular events.
• Researchers suggest the AI tool is a safe intervention that provides the added benefits of lower cost and greater sustainability.
• In the large study, the AI-based system improved stroke care and outcomes, supporting its potential as a scalable tool for routine stroke care, particularly in resource-limited settings.

Stroke is a significant global health concern and continues to be a leading cause of disability and death in the United States.

Evidence suggests that more than 795,000 people in the U.S. have a stroke each year, and nearly one in four of those are people who have had a previous stroke.

Clinicians play a critical role in preventing recurrent stroke. Typically, this occurs through implementing effective strategies, such as prevention plans, regular patient reviews, and addressing lifestyle modifications.

To assist with this, clinicians may consider clinical decision support systems (CDSS). These systems can help healthcare institutions analyze data from electronic health records and make recommendations to physicians by sending prompts and reminders in real-time

The potential scope of CDSS to help aid clinicians in complex decision-making processes for preventing stroke is increasing. However, many tools that utilize AI have not been rigorously evaluated, limiting their use.

Now, a large study published in The BMJ suggests an AI-powered CDSS may improve the quality of care and long-term outcomes for people who experience an acute ischemic stroke.

The findings suggest that such systems could offer a scalable and cost-effective way to enhance stroke management, particularly in regions with limited healthcare resources.

The use of AI technologies has increasingly been explored in healthcare, particularly for diagnosing disease, predicting outcomes, and supporting clinical decision making.

However, many AI tools designed for stroke care have not yet undergone rigorous evaluation in real-world clinical settings, limiting their widespread adoption.

To address this, researchers in China conducted a large trial to assess whether an AI-assisted CDSS could improve care quality and patient outcomes in routine practice.

The system analyzes brain scans to classify stroke causes and combines this with evidence-based treatment recommendations tailored to individual patients.

The research team suggests that the AI-based tool was associated with a significant reduction in subsequent vascular events compared with standard care.

Christopher Yi, MD, board certified vascular surgeon at MemorialCare Orange Coast Medical Center in Fountain Valley, CA, who was not involved in the study, suggests how AI could fit into stroke management.

“This study is the first of its kind to utilize AI for stroke care from being a diagnostic aid to being a tool that can improve care quality and reduce recurrent vascular events,” said Yi.

“In this study, the CDSS did more than read images: It integrated AI-assisted imaging, stroke-cause classification, reminders for needed evaluations, and guideline-based treatment recommendations,” he added.

“The biggest takeaway is that a well-integrated CDSS can help clinicians deliver more consistent evidence-based stroke care. It also helps guide interventionalists to better outcomes by improving stroke care quality and decreasing long term vascular events.”
– Christopher Yi, MD

The large study involved more than 21,000 participants with acute ischemic stroke admitted to 77 hospitals across China within 7 days of symptom onset. The individuals had an average age of 67, and just over one-third were female.

Between January 2021 and June 2023, 11,054 people received treatment at 38 hospitals supported by the AI-based CDSS. The other 10,549 participants at 39 hospitals received usual medical care.

Physicians in the intervention group were trained to use the system. The CDSS incorporated a range of patient-specific factors, including age, medical history, lifestyle, and hospital characteristics, when generating recommendations.

The study found that participants whose care was supported by the CDSS experienced fewer new vascular events at multiple follow-up points. This included recurrent stroke, heart attack, or related death.

At 3 months, 2.9% of those in the intervention group (320 of 11,054) experienced a new vascular event, compared with 3.9% in the control group (416 of 10,549), representing a 26% relative reduction.

This benefit persisted at 12 months, with event rates of 4% in the intervention group (440 of 11,054) versus 5.5% in the control group (576 of 10,549), representing a 27% reduction.

The research team also found that care quality measures were slightly higher in the intervention group, with performance scores of 91.4% compared with 89.8% in the usual care group.

Notably, the researchers add that the use of the AI system did not appear to increase risks. There were no significant differences between the groups in terms of disability, overall mortality, or bleeding complications at 3, 6, or 12 months.

When asked how clinically meaningful these improvements in care quality measures are, Yi told us: “Modest overall, but meaningful in the domains that matter most. The composite quality score improved from 89.8% to 91.4%, which by itself is not dramatic.”

“But several individual measures improved more substantially, including dual antiplatelet use, anticoagulation for atrial fibrillation, dysphagia screening, and DVT prophylaxis,” he noted. “Those are not trivial process metrics; they are directly tied to secondary prevention and complication avoidance.”

“The fact that recurrent vascular events fell from 3.9% to 2.9% at 3 months makes the quality gains feel clinically real rather than cosmetic,” Yi emphasized.

The authors note that the trial randomized hospitals rather than individual patients. This means that differences in care practices and follow-up outside the hospital could have influenced the results.

Despite this, the researchers emphasize that the system was easy to integrate into existing hospital infrastructure and required relatively minimal training.

“The biggest barriers are likely to be workflow integration, interoperability, imaging standardization, technical support, and clinician adoption,” Yi told Medical News Today.

“This system was integrated into the hospital information system, EMR [electronic medical record], and PACS [picture archiving and communication system], and physicians received training before rollout, which takes infrastructure and organizational commitment,” he continued.

“The paper also notes that hospitals already struggle with insufficient resources and heavy physician workloads, which are exactly the settings where implementation can be hardest even if the tool is potentially valuable,” said Yi.

“The next challenge is not proving that AI can help, but making it portable, explainable, affordable, and easy to trust across different practice environments,” he added.

The researchers suggest that AI-powered CDSS could serve as a comprehensive management tool, supporting both in-hospital care and secondary prevention strategies.

They add that it could represent a promising approach to delivering high-quality stroke care at scale, particularly in resource-constrained settings with a high burden of cerebrovascular disease.

As healthcare systems continue to explore the possible role of AI, studies like this indicate that such tools may deliver measurable benefits in real-world clinical practice.