AI to improve recognition of stroke symptoms in calls to emergency medical services

Jonathan Wenstrup, MD, PhD Candidate, Herlev Hospital, Herlev, Denmark, discusses his work looking at the use of artificial intelligence (AI) to support emergency call handlers in identifying stroke. Dr Wenstrup conducted a registry study assessing recognition of stroke symptoms by call handlers for the emergency medical call center and out-of-hours health service in Copenhagen, Denmark. It was shown that 64% of strokes were recognized by emergency medical call center handlers and 25% were recognized by out-of-hours health service call handlers. Notably, symptoms were more frequently overlooked in women, in those calling more than three hours from symptom onset, and for more severe strokes. Dr Wenstrup introduces his work using AI to try to enhance stroke recognition by call handlers and thus enable timely treatment initiation. His team designed a machine learning algorithm to analyze call data, detect patterns, and predict if a call was related to stroke or not. Having been trained on a data set of 1.5 million calls, the algorithm was tested on another data set of 344,000 calls, where it outperformed call handlers in both total stroke recognition and positive predictive value.
This work was conducted in collaboration with Corti, with funding from Trygfonden, The Capital Region of Denmark, and Region Zealand.
This interview took place at the 10th European Stroke Organisation Conference (ESOC) 2024 in Basel, Switzerland.
These works are owned by Magdalen Medical Publishing (MMP) and are protected by copyright laws and treaties around the world. All rights are reserved.