Fig. 6: Replication of feature performance in other data sets.
From: Metabolic phenotyping reveals an emerging role of ammonia abnormality in Alzheimer’s disease
Three-level meta-analysis forest plots of identified features on their alteration trends with AD progression (a; standardized mean difference with 95%CI of AD and CN stages; > 0, higher in AD; < 0, higher in CN) and on their associations with global cognition (b; Partial Spearman r value with 95%CI based on all available data. Age and sex were adjusted.). The p-values were based on two-sided tests. C-PAS (study No.9) and another 8 data sets derived from ADNI and ROSMAP were involved. 1: ADNI-Duke2016; 2: ADNI-Duke2017; 3: ADNI-California2017; 4: ADNI-Hawaii2021; 5: ADNI-Nightingale2021; 6: ADNI-DukeBAs2016; 7: ADNI-M2OVEAD2016; 8: Rosmap-Hawaii2017; 9: C-PAS-Shanghai2023. Variables were scaled to 0-1 within each study respectively and p-values were from random effect models. Cognition scores for ADNI, C-PAS, and ROSMAP data sets were ADAS-13, -1*MMSE, and a composite measure of global cognition created by averaging the z-scores of all tests respectively. Source data are provided as a Source Data file.
