MSR152: Natural History of Creutzfeldt-Jakob Disease: A Comparison of Manual Versus Manual plus AI-Assisted Approaches in Literature Review

Supplemental Methods and Results

METHODS

PubMed search strings:

("Creutzfeldt-Jakob Disease"[Title/Abstract] OR "CJD"[Title/Abstract])

AND ("natural history"[Title/Abstract] OR "disease progression"[Title/Abstract] OR "worsening"[Title/Abstract] OR "clinical course"[Title/Abstract] OR "longitudinal"[Title/Abstract])

NOT ("Animals"[MeSH Terms] OR "Animal Experimentation"[MeSH Terms] OR "Models, Animal"[MeSH Terms]) AND (english[Language]) AND (hasabstract[text])

Resulted in 51 articles with manual search using the terminology in blue for natural history

AI assist query for additional terminology for lit review suggested 10 more terms for disease progression (green, below) which helped identify 7 more articles. Of these 5 were case studies, one was focused on delirium and deemed irrelevant and only one deemed relevant which focused on diagnostic timelines of laboratory confirmed CJD cases.

PubMed search

(“Creutzfeldt-jakob Disease” [Title/Abstract] OR “CJD” [Title/Abstract]) AND (“natural history” [Title/Abstract] OR “disease progression” [Title/Abstract] OR “worsening” [Title/Abstract] OR “clinical course” [Title/Abstract] OR “longitudinal” [Title/Abstract] OR “time course” [Title/Abstract] OR “disease trajectory” [Title/Abstract] OR “clinical evolution” [Title/Abstract] OR “temporal pattern” [Title/Abstract] OR “disease timeline” [Title/Abstract] OR “progressive symptoms” [Title/Abstract] OR “onset to death” [Title/Abstract] OR “early signs” [Title/Abstract] OR “late-stage features” [Title/Abstract] OR “chronology of symptoms” [Title/Abstract]) NOT ("Animals"[MeSH Terms] OR "Animal Experimentation"[MeSH Terms] OR "Models, Animal"[MeSH Terms]) AND (english[Language]) AND (hasabstract[text])



Step 2 (Title and abstract screening) included a manual review of 58 abstracts. Of these, the majority (33) were case studies focusing on diagnostic challenges. Among the remaining 25 abstracts, 18 appeared relevant to the natural history of CJD, addressing aspects such as disease prognosis, phenotypic variations, genetics, and immune involvement, including 2 that focused on therapeutic research. Seven articles were deemed off-topic, as they focused on other disease states (ie, ALS, Alzheimer’s disease, delirium, dementia, BSE, and clinical guidelines). Of these, AI (through a program coded using Python), correctly flagged 4 as unimportant but incorrectly identified the remaining 3 as relevant.