Anoop Mayampurath1, Zahra Parnianpour2, Christopher T Richards3, William J Meurer4, Jungwha Lee5, Bruce Ankenman6, Ohad Perry6, Scott J Mendelson2, Jane L Holl2, Shyam Prabhakaran2. 1. Department of Pediatrics (A.M.), University of Chicago, IL. 2. Department of Neurology (Z.P., S.J.M., J.L.H., S.P.), University of Chicago, IL. 3. Department of Emergency Medicine, University of Cincinnati, OH (C.T.R.). 4. Department of Emergency Medicine, University of Michigan, Ann Arbor, IL (W.J.M.). 5. Department of Preventive Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL (J.L.). 6. Department of Industrial Engineering and Management Studies, Northwestern University (B.A., O.P.).
Abstract
Background and Purpose: Accurate prehospital diagnosis of stroke by emergency medical services (EMS) can increase treatments rates, mitigate disability, and reduce stroke deaths. We aimed to develop a model that utilizes natural language processing of EMS reports and machine learning to improve prehospital stroke identification. Methods: We conducted a retrospective study of patients transported by the Chicago EMS to 17 regional primary and comprehensive stroke centers. Patients who were suspected of stroke by the EMS or had hospital-diagnosed stroke were included in our cohort. Text within EMS reports were converted to unigram features, which were given as input to a support-vector machine classifier that was trained on 70% of the cohort and tested on the remaining 30%. Outcomes included final diagnosis of stroke versus nonstroke, large vessel occlusion, severe stroke (National Institutes of Health Stroke Scale score >5), and comprehensive stroke center-eligible stroke (large vessel occlusion or hemorrhagic stroke). Results: Of 965 patients, 580 (60%) had confirmed acute stroke. In a test set of 289 patients, the text-based model predicted stroke nominally better than models based on the Cincinnati Prehospital Stroke Scale (c-statistic: 0.73 versus 0.67, P=0.165) and was superior to the 3-Item Stroke Scale (c-statistic: 0.73 versus 0.53, P<0.001) scores. Improvements in discrimination were also observed for the other outcomes. Conclusions: We derived a model that utilizes clinical text from paramedic reports to identify stroke. Our results require validation but have the potential of improving prehospital routing protocols.
Background and Purpose: Accurate prehospital diagnosis of stroke by emergency medical services (EMS) can increase treatments rates, mitigate disability, and reduce stroke deaths. We aimed to develop a model that utilizes natural language processing of EMS reports and machine learning to improve prehospital stroke identification. Methods: We conducted a retrospective study of patients transported by the Chicago EMS to 17 regional primary and comprehensive stroke centers. Patients who were suspected of stroke by the EMS or had hospital-diagnosed stroke were included in our cohort. Text within EMS reports were converted to unigram features, which were given as input to a support-vector machine classifier that was trained on 70% of the cohort and tested on the remaining 30%. Outcomes included final diagnosis of stroke versus nonstroke, large vessel occlusion, severe stroke (National Institutes of Health Stroke Scale score >5), and comprehensive stroke center-eligible stroke (large vessel occlusion or hemorrhagic stroke). Results: Of 965 patients, 580 (60%) had confirmed acute stroke. In a test set of 289 patients, the text-based model predicted stroke nominally better than models based on the Cincinnati Prehospital Stroke Scale (c-statistic: 0.73 versus 0.67, P=0.165) and was superior to the 3-Item Stroke Scale (c-statistic: 0.73 versus 0.53, P<0.001) scores. Improvements in discrimination were also observed for the other outcomes. Conclusions: We derived a model that utilizes clinical text from paramedic reports to identify stroke. Our results require validation but have the potential of improving prehospital routing protocols.
Entities:
Keywords:
diagnosis; machine learning; natural language processing; patient; retrospective studies
Authors: Kennedy R Lees; Erich Bluhmki; Rüdiger von Kummer; Thomas G Brott; Danilo Toni; James C Grotta; Gregory W Albers; Markku Kaste; John R Marler; Scott A Hamilton; Barbara C Tilley; Stephen M Davis; Geoffrey A Donnan; Werner Hacke; Kathryn Allen; Jochen Mau; Dieter Meier; Gregory del Zoppo; D A De Silva; K S Butcher; M W Parsons; P A Barber; C Levi; C Bladin; G Byrnes Journal: Lancet Date: 2010-05-15 Impact factor: 79.321
Authors: Dariush Mozaffarian; Emelia J Benjamin; Alan S Go; Donna K Arnett; Michael J Blaha; Mary Cushman; Sarah de Ferranti; Jean-Pierre Després; Heather J Fullerton; Virginia J Howard; Mark D Huffman; Suzanne E Judd; Brett M Kissela; Daniel T Lackland; Judith H Lichtman; Lynda D Lisabeth; Simin Liu; Rachel H Mackey; David B Matchar; Darren K McGuire; Emile R Mohler; Claudia S Moy; Paul Muntner; Michael E Mussolino; Khurram Nasir; Robert W Neumar; Graham Nichol; Latha Palaniappan; Dilip K Pandey; Mathew J Reeves; Carlos J Rodriguez; Paul D Sorlie; Joel Stein; Amytis Towfighi; Tanya N Turan; Salim S Virani; Joshua Z Willey; Daniel Woo; Robert W Yeh; Melanie B Turner Journal: Circulation Date: 2014-12-17 Impact factor: 29.690
Authors: Oliver C Singer; Florian Dvorak; Richard du Mesnil de Rochemont; Heiner Lanfermann; Matthias Sitzer; Tobias Neumann-Haefelin Journal: Stroke Date: 2005-02-24 Impact factor: 7.914
Authors: Stephen W English; Alejandro A Rabinstein; Jay Mandrekar; James P Klaas Journal: J Stroke Cerebrovasc Dis Date: 2017-12-06 Impact factor: 2.136
Authors: Matthew I Miller; Agni Orfanoudaki; Michael Cronin; Hanife Saglam; Ivy So Yeon Kim; Oluwafemi Balogun; Maria Tzalidi; Kyriakos Vasilopoulos; Georgia Fanaropoulou; Nina M Fanaropoulou; Jack Kalin; Meghan Hutch; Brenton R Prescott; Benjamin Brush; Emelia J Benjamin; Min Shin; Asim Mian; David M Greer; Stelios M Smirnakis; Charlene J Ong Journal: Neurocrit Care Date: 2022-05-09 Impact factor: 3.532