Commentary
Video
Functionality of Integrating Artificial Intelligence Into Stroke Scales: Deepak K. Gulati, MD
Author(s):
The associate professor of neurology at The Ohio State University Wexner Medical Center provided insight on research assessing the use of a developed artificial intelligence/machine learning tool for automated stroke scale calculation. [WATCH TIME: 6 minutes]
WATCH TIME: 6 minutes
"All those components provide different results at the end of the day. We can provide 11 components, we can make 4 components. This is the same fundamental [approach] we can use to develop a stroke scale to be used in the field, as opposed to NIH-Stroke Scale."
An individual may be having a stroke if they demonstrate signs of sudden numbness or weakness in the face, arm, or leg, as well as if they display confusion, have trouble seeing, or are experiencing severe headache. In both pre-hospital, emergency room, in-hospital, and outpatient settings, there are various versions of stroke scales to determine a patient’s severity, but there are no automated tools for stroke scale calculation.
A new artificial intelligence (AI)/machine learning (ML)-based tool, developed by those at The Ohio State University Wexner Medical Center, uses audio/video interface to automate stroke scale calculation. This new type of approach can detect and track facial features and body parts to evaluate facial palsy and motor impairment of arms and legs, as well as verbal response for any speech or language impairment using different neural networks and AI-based algorithms. The speech system has been trained on over 30,000 tokens of speech to allow it to recognize symptoms of aphasia, a language disorder most often caused by stroke.
At the 2024 International Stroke Conference (ISC), held February 7-9, in Phoenix, Arizona, investigators presented a model assessing the AI/ML tool, with result showing a 91% test accuracy with facial asymmetry and motor weakness. Presented by lead investigator Deepak K. Gulati, MD, an associate professor of neurology at The Ohio State University Wexner Medical Center, the study underscored the promising diagnostic ability of the tool. Following the meeting, Gulati sat down with NeurologyLive® to discuss the clinical utility of the AI/MR-based approach and how it may improve stroke care. He gave an overview of the study, some of the greatest takeaways, and how the tool may be used going forward.
