Digitized and AI-Assisted Scoring of Cognitive Assessments in ALS and Neurodegenerative Diseases: A Scoping Review of Consistency, Reliability, and Cross-Site Applicability

Research overview

The text below is the source article’s original abstract under a verified reuse license. cc by

Background. Standardized cognitive assessment across clinical sites remains a major challenge in amyotrophic lateral sclerosis (ALS) and related neurodegenerative diseases. The Edinburgh Cognitive and Behavioral ALS Screen (ECAS) is the most validated and widely-used ALS-specific cognitive assessment tool, yet paper-and-pencil manual scoring introduces inter-rater variability that can significantly limit reliability and reproducibility in multi-site clinical trials. Digitized administration and artificial intelligence-assisted scoring may improve consistency and scalability, but the current landscape across ALS and related diseases has not been systematically mapped.Objective. To map existing evidence on digitized and AI-assisted scoring of cognitive assessments in ALS and other neurodegenerative and neuroinflammatory diseases, with emphasis on inter-rater reliability and cross-site consistency.Methods. We conducted a comprehensive scoping review following Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines. Eligible studies evaluated neuropsychological or cognitive assessment tools using automated, computerized, digital, or AI-assisted scoring in ALS and related conditions, including Alzheimer's disease, Parkinson's disease, multiple sclerosis, dementia, and neuroinflammatory disease. An initial broad PubMed search was refined to focus on records explicitly addressing automated or AI-assisted scoring in titles and abstracts. Records were loaded into Rayyan for independent blinded review by three reviewers, with discrepancies resolved through discussion until consensus was reached. Data were extracted using standardized templates.Results. The initial search identified 1,503 records and was refined to 168 records. After duplicate removal, 166 records underwent title, abstract, and full-text screening against predefined inclusion and exclusion criteria. Seven primary studies met inclusion criteria: two in multiple sclerosis, two in the Alzheimer's disease continuum, and three in dementia or mild cognitive impairment. AI and digital approaches included tablet-based digitization with automated scoring, convolutional neural networks, and large language models. All seven studies reported high agreement with human or paper-based reference standards. No primary study evaluating AI-assisted scoring of the ECAS or ALS-specific cognitive screens was identified.Conclusions. This review maps AI-assisted and digitized cognitive assessment scoring in neurodegenerative and neuroinflammatory diseases. Findings reveal a critical gap in ALS-specific cognitive tools and ECAS scoring. Methodological approaches from related diseases, especially automated scoring of verbal and written responses, may inform future development of standardized AI-assisted ECAS protocols. Such tools could improve consistency, reliability, and scalability across multi-site ALS clinical trials.

Related topics