Hooman Kamel1, Babak B Navi1, Neal S Parikh1, Alexander E Merkler1, Peter M Okin2, Richard B Devereux2, Jonathan W Weinsaft2, Jiwon Kim2, Jim W Cheung2, Luke K Kim2, Barbara Casadei3, Costantino Iadecola1, Mert R Sabuncu4, Ajay Gupta5, Iván Díaz6. 1. Clinical and Translational Neuroscience Unit, Department of Neurology, Feil Family Brain and Mind Research Institute (H.K., B.B.N., N.S.P., A.E.M., C.I.), Weill Cornell Medical College, New York. 2. Division of Cardiology (P.M.O., R.B.D., J.W.W., J.K., J.W.C., L.K.K.), Weill Cornell Medical College, New York. 3. Division of Cardiovascular Medicine, University of Oxford, United Kingdom (B.C.). 4. School of Electrical and Computer Engineering, and Meinig School of Biomedical Engineering, Cornell University, Ithaca, NY (M.R.S.). 5. Department of Radiology (A.G.), Weill Cornell Medical College, New York. 6. Division of Biostatistics and Epidemiology (I.D.), Weill Cornell Medical College, New York.
Abstract
BACKGROUND AND PURPOSE: One-fifth of ischemic strokes are embolic strokes of undetermined source (ESUS). Their theoretical causes can be classified as cardioembolic versus noncardioembolic. This distinction has important implications, but the categories' proportions are unknown. METHODS: Using data from the Cornell Acute Stroke Academic Registry, we trained a machine-learning algorithm to distinguish cardioembolic versus non-cardioembolic strokes, then applied the algorithm to ESUS cases to determine the predicted proportion with an occult cardioembolic source. A panel of neurologists adjudicated stroke etiologies using standard criteria. We trained a machine learning classifier using data on demographics, comorbidities, vitals, laboratory results, and echocardiograms. An ensemble predictive method including L1 regularization, gradient-boosted decision tree ensemble (XGBoost), random forests, and multivariate adaptive splines was used. Random search and cross-validation were used to tune hyperparameters. Model performance was assessed using cross-validation among cases of known etiology. We applied the final algorithm to an independent set of ESUS cases to determine the predicted mechanism (cardioembolic or not). To assess our classifier's validity, we correlated the predicted probability of a cardioembolic source with the eventual post-ESUS diagnosis of atrial fibrillation. RESULTS: Among 1083 strokes with known etiologies, our classifier distinguished cardioembolic versus noncardioembolic cases with excellent accuracy (area under the curve, 0.85). Applied to 580 ESUS cases, the classifier predicted that 44% (95% credibility interval, 39%-49%) resulted from cardiac embolism. Individual ESUS patients' predicted likelihood of cardiac embolism was associated with eventual atrial fibrillation detection (OR per 10% increase, 1.27 [95% CI, 1.03-1.57]; c-statistic, 0.68 [95% CI, 0.58-0.78]). ESUS patients with high predicted probability of cardiac embolism were older and had more coronary and peripheral vascular disease, lower ejection fractions, larger left atria, lower blood pressures, and higher creatinine levels. CONCLUSIONS: A machine learning estimator that distinguished known cardioembolic versus noncardioembolic strokes indirectly estimated that 44% of ESUS cases were cardioembolic.
BACKGROUND AND PURPOSE: One-fifth of ischemic strokes are embolic strokes of undetermined source (ESUS). Their theoretical causes can be classified as cardioembolic versus noncardioembolic. This distinction has important implications, but the categories' proportions are unknown. METHODS: Using data from the Cornell Acute Stroke Academic Registry, we trained a machine-learning algorithm to distinguish cardioembolic versus non-cardioembolic strokes, then applied the algorithm to ESUS cases to determine the predicted proportion with an occult cardioembolic source. A panel of neurologists adjudicated stroke etiologies using standard criteria. We trained a machine learning classifier using data on demographics, comorbidities, vitals, laboratory results, and echocardiograms. An ensemble predictive method including L1 regularization, gradient-boosted decision tree ensemble (XGBoost), random forests, and multivariate adaptive splines was used. Random search and cross-validation were used to tune hyperparameters. Model performance was assessed using cross-validation among cases of known etiology. We applied the final algorithm to an independent set of ESUS cases to determine the predicted mechanism (cardioembolic or not). To assess our classifier's validity, we correlated the predicted probability of a cardioembolic source with the eventual post-ESUS diagnosis of atrial fibrillation. RESULTS: Among 1083 strokes with known etiologies, our classifier distinguished cardioembolic versus noncardioembolic cases with excellent accuracy (area under the curve, 0.85). Applied to 580 ESUS cases, the classifier predicted that 44% (95% credibility interval, 39%-49%) resulted from cardiac embolism. Individual ESUS patients' predicted likelihood of cardiac embolism was associated with eventual atrial fibrillation detection (OR per 10% increase, 1.27 [95% CI, 1.03-1.57]; c-statistic, 0.68 [95% CI, 0.58-0.78]). ESUS patients with high predicted probability of cardiac embolism were older and had more coronary and peripheral vascular disease, lower ejection fractions, larger left atria, lower blood pressures, and higher creatinine levels. CONCLUSIONS: A machine learning estimator that distinguished known cardioembolic versus noncardioembolic strokes indirectly estimated that 44% of ESUS cases were cardioembolic.
Authors: Hooman Kamel; W T Longstreth; David L Tirschwell; Richard A Kronmal; Joseph P Broderick; Yuko Y Palesch; Caitlyn Meinzer; Catherine Dillon; Irene Ewing; Judith A Spilker; Marco R Di Tullio; Eldad A Hod; Elsayed Z Soliman; Seemant Chaturvedi; Claudia S Moy; Scott Janis; Mitchell Sv Elkind Journal: Int J Stroke Date: 2018-09-10 Impact factor: 5.266
Authors: Tobias Geisler; Sven Poli; Christoph Meisner; Juergen Schreieck; Christine S Zuern; Thomas Nägele; Johannes Brachmann; Werner Jung; Georg Gahn; Elisabeth Schmid; Hansjörg Bäezner; Timea Keller; Gabor C Petzold; Jan-Wilko Schrickel; Jan Liman; Rolf Wachter; Frauke Schön; Martin Schabet; Alfred Lindner; Albert C Ludolph; Hubert Kimmig; Sebastian Jander; Uwe Schlegel; Meinrad Gawaz; Ulf Ziemann Journal: Int J Stroke Date: 2016-11-23 Impact factor: 5.266
Authors: Hans-Christoph Diener; Ralph L Sacco; J Donald Easton; Christopher B Granger; Richard A Bernstein; Shinichiro Uchiyama; Jörg Kreuzer; Lisa Cronin; Daniel Cotton; Claudia Grauer; Martina Brueckmann; Marina Chernyatina; Geoffrey Donnan; José M Ferro; Martin Grond; Bernd Kallmünzer; Jerzy Krupinski; Byung-Chul Lee; Robin Lemmens; Jaime Masjuan; Miroslav Odinak; Jeffrey L Saver; Peter D Schellinger; Danilo Toni; Kazunori Toyoda Journal: N Engl J Med Date: 2019-05-16 Impact factor: 91.245
Authors: Romain Pirracchio; Maya L Petersen; Marco Carone; Matthieu Resche Rigon; Sylvie Chevret; Mark J van der Laan Journal: Lancet Respir Med Date: 2014-11-24 Impact factor: 30.700
Authors: Tobias M Freilinger; Andreas Schindler; Caroline Schmidt; Jochen Grimm; Clemens Cyran; Florian Schwarz; Fabian Bamberg; Jennifer Linn; Maximilian Reiser; Chun Yuan; Konstantin Nikolaou; Martin Dichgans; Tobias Saam Journal: JACC Cardiovasc Imaging Date: 2012-04
Authors: Rafael Lozano; Mohsen Naghavi; Kyle Foreman; Stephen Lim; Kenji Shibuya; Victor Aboyans; Jerry Abraham; Timothy Adair; Rakesh Aggarwal; Stephanie Y Ahn; Miriam Alvarado; H Ross Anderson; Laurie M Anderson; Kathryn G Andrews; Charles Atkinson; Larry M Baddour; Suzanne Barker-Collo; David H Bartels; Michelle L Bell; Emelia J Benjamin; Derrick Bennett; Kavi Bhalla; Boris Bikbov; Aref Bin Abdulhak; Gretchen Birbeck; Fiona Blyth; Ian Bolliger; Soufiane Boufous; Chiara Bucello; Michael Burch; Peter Burney; Jonathan Carapetis; Honglei Chen; David Chou; Sumeet S Chugh; Luc E Coffeng; Steven D Colan; Samantha Colquhoun; K Ellicott Colson; John Condon; Myles D Connor; Leslie T Cooper; Matthew Corriere; Monica Cortinovis; Karen Courville de Vaccaro; William Couser; Benjamin C Cowie; Michael H Criqui; Marita Cross; Kaustubh C Dabhadkar; Nabila Dahodwala; Diego De Leo; Louisa Degenhardt; Allyne Delossantos; Julie Denenberg; Don C Des Jarlais; Samath D Dharmaratne; E Ray Dorsey; Tim Driscoll; Herbert Duber; Beth Ebel; Patricia J Erwin; Patricia Espindola; Majid Ezzati; Valery Feigin; Abraham D Flaxman; Mohammad H Forouzanfar; Francis Gerry R Fowkes; Richard Franklin; Marlene Fransen; Michael K Freeman; Sherine E Gabriel; Emmanuela Gakidou; Flavio Gaspari; Richard F Gillum; Diego Gonzalez-Medina; Yara A Halasa; Diana Haring; James E Harrison; Rasmus Havmoeller; Roderick J Hay; Bruno Hoen; Peter J Hotez; Damian Hoy; Kathryn H Jacobsen; Spencer L James; Rashmi Jasrasaria; Sudha Jayaraman; Nicole Johns; Ganesan Karthikeyan; Nicholas Kassebaum; Andre Keren; Jon-Paul Khoo; Lisa Marie Knowlton; Olive Kobusingye; Adofo Koranteng; Rita Krishnamurthi; Michael Lipnick; Steven E Lipshultz; Summer Lockett Ohno; Jacqueline Mabweijano; Michael F MacIntyre; Leslie Mallinger; Lyn March; Guy B Marks; Robin Marks; Akira Matsumori; Richard Matzopoulos; Bongani M Mayosi; John H McAnulty; Mary M McDermott; John McGrath; George A Mensah; Tony R Merriman; Catherine Michaud; Matthew Miller; Ted R Miller; Charles Mock; Ana Olga Mocumbi; Ali A Mokdad; Andrew Moran; Kim Mulholland; M Nathan Nair; Luigi Naldi; K M Venkat Narayan; Kiumarss Nasseri; Paul Norman; Martin O'Donnell; Saad B Omer; Katrina Ortblad; Richard Osborne; Doruk Ozgediz; Bishnu Pahari; Jeyaraj Durai Pandian; Andrea Panozo Rivero; Rogelio Perez Padilla; Fernando Perez-Ruiz; Norberto Perico; David Phillips; Kelsey Pierce; C Arden Pope; Esteban Porrini; Farshad Pourmalek; Murugesan Raju; Dharani Ranganathan; Jürgen T Rehm; David B Rein; Guiseppe Remuzzi; Frederick P Rivara; Thomas Roberts; Felipe Rodriguez De León; Lisa C Rosenfeld; Lesley Rushton; Ralph L Sacco; Joshua A Salomon; Uchechukwu Sampson; Ella Sanman; David C Schwebel; Maria Segui-Gomez; Donald S Shepard; David Singh; Jessica Singleton; Karen Sliwa; Emma Smith; Andrew Steer; Jennifer A Taylor; Bernadette Thomas; Imad M Tleyjeh; Jeffrey A Towbin; Thomas Truelsen; Eduardo A Undurraga; N Venketasubramanian; Lakshmi Vijayakumar; Theo Vos; Gregory R Wagner; Mengru Wang; Wenzhi Wang; Kerrianne Watt; Martin A Weinstock; Robert Weintraub; James D Wilkinson; Anthony D Woolf; Sarah Wulf; Pon-Hsiu Yeh; Paul Yip; Azadeh Zabetian; Zhi-Jie Zheng; Alan D Lopez; Christopher J L Murray; Mohammad A AlMazroa; Ziad A Memish Journal: Lancet Date: 2012-12-15 Impact factor: 79.321
Authors: Brian Mac Grory; Shadi Yaghi; Charlotte Cordonnier; Luciano A Sposato; Jose G Romano; Seemant Chaturvedi Journal: Circ Res Date: 2022-04-14 Impact factor: 23.213