Tau-CTFP: A connectome-informed continuous-time first-passage model for forecasting the onset of regional tau PET abnormality

Tau-CTFP forecasts first regional tau PET abnormality in amyloid-positive ADNI participants, with top-ranked precision of 59.5% across 168 follow-up visits.

• Why it matters: Tau PET burden tracks later cognitive decline, but clinicians and trials need individualized, horizon-specific predictions of where abnormality first appears.
• What they did: The connectome-informed generative continuous-time first-passage model uses normative tau scores, network exposure to previously abnormal regions, and learned regional susceptibility, trained on pooled consecutive scan pairs.
• The result: In participant-separated ranking evaluation, mean average precision was 0.584 (95% CI 0.536–0.632) and AUROC was 0.820 (95% CI 0.793–0.847), with dependence improving onset-count uncertainty versus independence.

AI summary of the abstract · may contain errors · not medical advice · read the paper.