The Cognitive Makeup of Writing: Multivariate Analysis of Writing Impairments Following Stroke

Abstract Writing is a recently acquired skill to human behavioral repertoire, essential in industrialized societies. In the clinic, writing impairment evident one-third of stroke patients. This study aimed find out cognitive features that contribute patients two different systems (logographic and phonological). Cognitive profiles were assessed using Birmingham Screen cohorts, China (244 patients) UK (501 patients). The datasets analyzed separately an identical procedure. Elastic net was used rank importance abilities (features) skill; linear support vector machine identify discriminative needed accurately with without impairments. prediction performance evaluated area under curve (AUC), accuracy (ACC), sensitivity (SEN), specificity (SPE). For cohort, numbers, complex figure copy, number calculation obtained good on impairments AUC 0.85 ± 0.06, ACC (89 3) %, SEN (81 10) SPE (90 27) %. Concerning data, calculation, non-word reading, auditory sustain attention achieved 0.79 0.04, (83 (74 9) (84 A small both cohorts (China: 9/69, UK: 24/137), who impaired writing, consistently misclassified. Two patients, one each showed selective while all remaining attention, language, and/or praxis tasks. results capability write numbers manipulate them critical for predicting across systems. Reading not predictor cohorts. Constructive (measured by copy) relevant classification characters-based (China), phonological reading) important alphabetic (UK). proportion minority cases deficits related profiles. findings this highlight multifaceted nature potential use computation methods revealing hidden structures neuropsychological research.