JMIR Publications

ABSTRACT

Alzheimer's Disease (AD) is diagnosed by categorizing individuals based on the severity of cognitive impairment. This systematic review and meta-analysis was aimed to assess AD prevalence across different stages utilizing machine learning (ML) approaches. Forest plots were used to depict the prevalence of AD subjects at different stages. According to the PRISMA 2020 guidelines, articles were selected in three phases: identification, screening, and final inclusion. For the final analysis, 24 articles that met the selection criteria were identified. ML approaches for AD diagnosis were selected based on their relevance to the investigation. Six studies focused on two classes of patients such as Cognitively Normal (CN) and AD. They reported a pooled prevalence of 49.28% (95% CI: 46.12-52.45%, p=0.32). The prevalence estimate for the three stages of cognitive impairment (CN, (Mild Cognitive Impairment (MCI), and AD) was 29.75% (95% CI: 25.11-34.84%, p<0.01). Among five studies with 14,839 subjects, the analysis of four stages [i.e., Non demented (ND), Moderatly demented (MoD), Mild demented (MD), and AD] found an overall prevalence of 13.13% (95% CI: 3.75-36.66%, p<0.01). Four studies involving 3,819 subjects estimated the prevalence of six stages [CN, Significant Memory Concern (SMC), Early Mild Cognitive Impairment (EMCI), MCI, Late Mild Cognitive Impairment (LMCI), and AD], yielding a prevalence of 23.75% (95% CI: 12.22-41.12%, p<0.01). AD prevalence estimates are influenced by demographic and setting characteristics, as evidenced by the substantial heterogeneity observed across studies. The above findings suggest the usefulness of ML approaches in describing AD prevalence across various stages, providing valuable insights for future research.