Gabrielle Bunney1, Julianne Murphy2, Katharine Colton3, Hanyin Wang4, Hye Jung Shin5, Roland Faigle6, Andrew M Naidech3,5. 1. Department of Emergency Medicine, Northwestern University, 625 N Michigan Ave Suite 1150, Chicago, IL, 60611, USA. gabrielle.bunney@northwestern.edu. 2. Center for Education in Health Sciences, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA. 3. Department of Neurology, Northwestern University, Chicago, IL, USA. 4. Driskill Graduate School of Life Sciences, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA. 5. Institute for Public Health and Medicine, Northwestern University, Chicago, IL, USA. 6. Department of Neurology, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
Abstract
BACKGROUND: Seizures are a harmful complication of acute intracerebral hemorrhage (ICH). "Early" seizures in the first week after ICH are a risk factor for deterioration, later seizures, and herniation. Ideally, seizure medications after ICH would only be administered to patients with a high likelihood to have seizures. We developed and validated machine learning (ML) models to predict early seizures after ICH. METHODS: We used two large datasets to train and then validate our models in an entirely independent test set. The first model ("CAV") predicted early seizures from a subset of variables of the CAVE score (a prediction rule for later seizures)-cortical hematoma location, age less than 65 years, and hematoma volume greater than 10 mL-whereas early seizure was the dependent variable. We attempted to improve on the "CAV" model by adding anticoagulant use, antiplatelet use, Glasgow Coma Scale, international normalized ratio, and systolic blood pressure ("CAV + "). For each model we used logistic regression, lasso regression, support vector machines, boosted trees (Xgboost), and random forest models. Final model performance was reported as the area under the receiver operating characteristic curve (AUC) using receiver operating characteristic models for the test data. The setting of the study was two large academic institutions: institution 1, 634 patients; institution 2, 230 patients. There were no interventions. RESULTS: Early seizures were predicted across the ML models by the CAV score in test data, (AUC 0.72, 95% confidence interval 0.62-0.82). The ML model that predicted early seizure better in the test data was Xgboost (AUC 0.79, 95% confidence interval 0.71-0.87, p = 0.04) compared with the CAV model AUC. CONCLUSIONS: Early seizures after ICH are predictable. Models using cortical hematoma location, age less than 65 years, and hematoma volume greater than 10 mL had a good accuracy rate, and performance improved with more independent variables. Additional methods to predict seizures could improve patient selection for monitoring and prophylactic seizure medications.
BACKGROUND: Seizures are a harmful complication of acute intracerebral hemorrhage (ICH). "Early" seizures in the first week after ICH are a risk factor for deterioration, later seizures, and herniation. Ideally, seizure medications after ICH would only be administered to patients with a high likelihood to have seizures. We developed and validated machine learning (ML) models to predict early seizures after ICH. METHODS: We used two large datasets to train and then validate our models in an entirely independent test set. The first model ("CAV") predicted early seizures from a subset of variables of the CAVE score (a prediction rule for later seizures)-cortical hematoma location, age less than 65 years, and hematoma volume greater than 10 mL-whereas early seizure was the dependent variable. We attempted to improve on the "CAV" model by adding anticoagulant use, antiplatelet use, Glasgow Coma Scale, international normalized ratio, and systolic blood pressure ("CAV + "). For each model we used logistic regression, lasso regression, support vector machines, boosted trees (Xgboost), and random forest models. Final model performance was reported as the area under the receiver operating characteristic curve (AUC) using receiver operating characteristic models for the test data. The setting of the study was two large academic institutions: institution 1, 634 patients; institution 2, 230 patients. There were no interventions. RESULTS: Early seizures were predicted across the ML models by the CAV score in test data, (AUC 0.72, 95% confidence interval 0.62-0.82). The ML model that predicted early seizure better in the test data was Xgboost (AUC 0.79, 95% confidence interval 0.71-0.87, p = 0.04) compared with the CAV model AUC. CONCLUSIONS: Early seizures after ICH are predictable. Models using cortical hematoma location, age less than 65 years, and hematoma volume greater than 10 mL had a good accuracy rate, and performance improved with more independent variables. Additional methods to predict seizures could improve patient selection for monitoring and prophylactic seizure medications.
Authors: Andrew M Naidech; Jennifer Beaumont; Kathryn Muldoon; Eric M Liotta; Matthew B Maas; Matthew B Potts; Babak S Jahromi; David Cella; Shyam Prabhakaran; Jane L Holl Journal: Crit Care Med Date: 2018-09 Impact factor: 7.598
Authors: P M Vespa; K O'Phelan; M Shah; J Mirabelli; S Starkman; C Kidwell; J Saver; M R Nuwer; J G Frazee; D A McArthur; N A Martin Journal: Neurology Date: 2003-05-13 Impact factor: 9.910
Authors: Elena Haapaniemi; Daniel Strbian; Costanza Rossi; Jukka Putaala; Tuulia Sipi; Satu Mustanoja; Tiina Sairanen; Sami Curtze; Jarno Satopää; Reina Roivainen; Markku Kaste; Charlotte Cordonnier; Turgut Tatlisumak; Atte Meretoja Journal: Stroke Date: 2014-05-29 Impact factor: 7.914
Authors: Steven R Messé; Lauren H Sansing; Brett L Cucchiara; Susan T Herman; Patrick D Lyden; Scott E Kasner Journal: Neurocrit Care Date: 2009-03-25 Impact factor: 3.210
Authors: Zubeda B Sheikh; Christoph Stretz; Carolina B Maciel; Monica B Dhakar; Hailey Orgass; Ognen A Petroff; Lawrence J Hirsch; Emily J Gilmore Journal: Crit Care Med Date: 2020-06 Impact factor: 7.598
Authors: Andrew M Naidech; Rajeev K Garg; Storm Liebling; Kimberly Levasseur; Micheal P Macken; Stephan U Schuele; H Hunt Batjer Journal: Stroke Date: 2009-09-24 Impact factor: 7.914
Authors: Salim S Virani; Alvaro Alonso; Hugo J Aparicio; Emelia J Benjamin; Marcio S Bittencourt; Clifton W Callaway; April P Carson; Alanna M Chamberlain; Susan Cheng; Francesca N Delling; Mitchell S V Elkind; Kelly R Evenson; Jane F Ferguson; Deepak K Gupta; Sadiya S Khan; Brett M Kissela; Kristen L Knutson; Chong D Lee; Tené T Lewis; Junxiu Liu; Matthew Shane Loop; Pamela L Lutsey; Jun Ma; Jason Mackey; Seth S Martin; David B Matchar; Michael E Mussolino; Sankar D Navaneethan; Amanda Marma Perak; Gregory A Roth; Zainab Samad; Gary M Satou; Emily B Schroeder; Svati H Shah; Christina M Shay; Andrew Stokes; Lisa B VanWagner; Nae-Yuh Wang; Connie W Tsao Journal: Circulation Date: 2021-01-27 Impact factor: 29.690
Authors: Paul M Vespa; DaiWai M Olson; Sayona John; Kyle S Hobbs; Kapil Gururangan; Kun Nie; Masoom J Desai; Matthew Markert; Josef Parvizi; Thomas P Bleck; Lawrence J Hirsch; M Brandon Westover Journal: Crit Care Med Date: 2020-09 Impact factor: 9.296
Authors: Andrew M Naidech; Bradley Weaver; Matthew Maas; Thomas P Bleck; Stephen VanHaerents; Stephan U Schuele Journal: Crit Care Med Date: 2021-06-01 Impact factor: 9.296