Fig. 4: Prognostic performance of tau-PET in individuals screened-in versus screened-out based on plasma pTau217 levels.

This figure shows the added prognostic value of tau-PET in predicting annual change in MMSE separately for individuals below (left panels; dark green) and above the referral cutoff (right panels; orange), based on the 95% sensitivity plasma pTau217 cutoff herein calculated, compared to a demographic model and a plasma pTau217 model. Each of the four models are represented in the y-axis, and the x-axis represents the adjusted R² of the models, indicating how well they predict annual change in MMSE. The weighted Akaike Information Criterion (wAIC), a measure from 0 to 1 with a higher value indicating which model is more likely to be correct among those tested, is indicated in vertically disposed black text. In both SCD-MCI (A) and all-cause dementia (B) clinical populations, a tau-PET scan was not useful for screened-out individuals (SCD-MCI: n = 147; All-cause dementia: n = 30), with the demographic model being more parsimonious; for those above the referral cutoff (SCD-MCI: n = 101; All-cause dementia: n = 123), the tau-PET-only model was significantly better than the other three models. Source data are provided as a Source Data File. PET positron emission tomography, pTau phosphorylated tau, R2 coefficient of determination, wAIC weighted Akaike Information Criterion, MMSE Mini-Mental State Exam, MCI mild cognitive impairment, SCD subjective cognitive decline.