Prediction of mild cognitive impairment progression using time-sensitive multimodal biomarkers

Alzheimers disease (AD) develops over a prolonged asymptomatic phase marked by silent pathology. Identifying cognitively unimpaired individuals likely to progress to mild cognitive impairment (MCI) is essential for early intervention. We investigated whether multimodal combinations of biomarkers, including frequency-specific neurophysiological activity, enhance prediction beyond demographic and genetic factors in older adults (n = 102; 31 progressors; mean follow-up = 5.9 years). Biomarkers included MEG-derived alpha power, MRI-derived hippocampal volume, plasma A{beta}42/40 ratio and p-tau217, and neocortical A{beta} and entorhinal tau PET. Cox regression models estimated progression risk and tested time-varying prognostic effects. Neurophysiological and proteinopathy biomarkers improved prediction beyond clinical and genetic factors. Elevated alpha power predicted short-term risk, but its predictive value weakened over time, whereas high neocortical A{beta} became increasingly predictive with longer follow-up. Plasma A{beta}42/40, p-tau217, and tau PET each conferred higher risk, while hippocampal volume did not. Findings support a multimodal, time-sensitive framework for individualized risk prediction in preclinical AD.