Machine learning for amyloid and tau pathology prediction

Method and apparatus for predicting amyloid beta (Aβ) and phosphorylated tau (p-tau) biomarker levels in the cerebrospinal fluid (CSF) of patients. Embodiments include determining current values for a plurality of easily-measurable attributes of a first patient. Embodiments include analyzing data associated with a cohort of patients having known measurements of Aβ and p-tau biomarker levels, including determined values for the plurality of easily-measureable attributes. Embodiments include generating a predicted value for Aβ and/or p-tau biomarker levels for the first patient. Embodiments include generating a risk of the first patient developing AD at a future time, generating a probability of a patient's predicted rate of decline, and/or generating a probability of a patient's age at the onset of dementia, based on the predicted values for Aβ and/or p-tau biomarker levels.