Researchers from the Nutritional Immunology and Molecular Medicine Laboratory and the Department of Biomedical and Translational Informatics at Geisinger Health System teamed up to develop an artificial neural network (ANN) model to accurately recognize stoke symptoms.
The artificial intelligence (AI) model utilizes millions of data points provided by Geisinger Health System to precisely identify, prevent and treat stroke based on a patient's personalized health record. By accurately identifying stoke patients as the enter the emergency room, the model has the potential to be a new data-driven method of improving patient outcomes.
“Advanced machine-learning methods will be driving the next generation of personalized medicine at the clinical and genomic levels; however, these methods and their outcomes will have an added value if we let models actively learn from experts and experts learn from models. Our team has applied AI successfully to develop a data-driven triage process for classifying stroke patients. Ongoing collaborative studies are also applying these same AI methods successfully in infectious and immune-mediated diseases,” said Vida Abedi, a researcher at Geisinger Health System and adjunct faculty member in NIMML.
The study enrolled 260 patients who visited the emergency room with stroke-like symptoms, within 4.5 hours of onset. The ANN model showed 80 percent sensitivity and 86.2 percent specificity, proving an effective tool in identify stroke patients.