Machine Learning Methods Predict Individual Upper-Limb Motor Impairment Following Therapy in Chronic Stroke.

TitleMachine Learning Methods Predict Individual Upper-Limb Motor Impairment Following Therapy in Chronic Stroke.
Publication TypeJournal Article
Year of Publication2020
AuthorsTozlu C, Edwards D, Boes A, Labar D, K Tsagaris Z, Silverstein J, Lane HPepper, Sabuncu MR, Liu C, Kuceyeski A
JournalNeurorehabil Neural Repair
Volume34
Issue5
Pagination428-439
Date Published2020 05
ISSN1552-6844
KeywordsAged, Chronic Disease, Evoked Potentials, Motor, Exercise Therapy, Female, Humans, Machine Learning, Magnetic Resonance Imaging, Male, Middle Aged, Motor Cortex, Neural Networks, Computer, Outcome Assessment, Health Care, Severity of Illness Index, Stroke, Stroke Rehabilitation, Support Vector Machine, Transcranial Magnetic Stimulation, Upper Extremity
Abstract

. Accurate prediction of clinical impairment in upper-extremity motor function following therapy in chronic stroke patients is a difficult task for clinicians but is key in prescribing appropriate therapeutic strategies. Machine learning is a highly promising avenue with which to improve prediction accuracy in clinical practice. . The objective was to evaluate the performance of 5 machine learning methods in predicting postintervention upper-extremity motor impairment in chronic stroke patients using demographic, clinical, neurophysiological, and imaging input variables. . A total of 102 patients (female: 31%, age 61 ± 11 years) were included. The upper-extremity Fugl-Meyer Assessment (UE-FMA) was used to assess motor impairment of the upper limb before and after intervention. Elastic net (EN), support vector machines, artificial neural networks, classification and regression trees, and random forest were used to predict postintervention UE-FMA. The performances of methods were compared using cross-validated . . EN performed significantly better than other methods in predicting postintervention UE-FMA using demographic and baseline clinical data (median < .05). Preintervention UE-FMA and the difference in motor threshold (MT) between the affected and unaffected hemispheres were the strongest predictors. The difference in MT had greater importance than the absence or presence of a motor-evoked potential (MEP) in the affected hemisphere. . Machine learning methods may enable clinicians to accurately predict a chronic stroke patient's postintervention UE-FMA. Interhemispheric difference in the MT is an important predictor of chronic stroke patients' response to therapy and, therefore, could be included in prospective studies.

DOI10.1177/1545968320909796
Alternate JournalNeurorehabil Neural Repair
PubMed ID32193984
PubMed Central IDPMC7217740
Grant ListR01 AG053949 / AG / NIA NIH HHS / United States
R01 LM012719 / LM / NLM NIH HHS / United States
R01 NS102646 / NS / NINDS NIH HHS / United States
R21 NS104634 / NS / NINDS NIH HHS / United States
Related Institute: 
Brain Health Imaging Institute (BHII)

Weill Cornell Medicine
Department of Radiology
525 East 68th Street New York, NY 10065