Classification of tau status with machine learning models in amyloid-positive cohorts.

AI / informaticsAlzheimer's & dementia : the journal of the Alzheimer's Association· Aug 2026

Classification of tau status with machine learning models in amyloid-positive cohorts.

Yun-Chi Lin, Sheng-Chieh Chiu, Rebecca Massey … Alzheimer's Disease Neuroimaging Initiative

Read paper

Nuclear MedicineAI / informaticsDiagnosisQ1IF 11.9

Machine learning model using magnetic resonance imaging and amyloid positron emission tomography features accurately classified tau positivity.

Classification of tau status with machine learning models in amyloid-positive cohorts.

Yun-Chi Lin … Alzheimer's Disease Neuroimaging Initiative

Alzheimer's & dementia · Aug 2026

Background

Although tau positron emission tomography (positron emission tomography) imaging is effective for staging tau pathology, it is limited clinically by cost and availability.

Methods

Multiple machine learning models were developed to classify tau positivity in the Braak III/IV region using structural magnetic resonance imaging, amyloid positron emission tomography, and demographic features.

n = 410

Results

0.92

model correctly told tau-positive from tau-negative brains almost every time

n = 410

More results

Combined external validation yielded accuracy/sensitivity/specificity of 85%/83%/85%.

Subjects with mild cognitive impairment and predicted tau positivity progressed to AD at a significantly faster pace (p < 10 -6 ).

“

Conclusion

DISCUSSION: Our model demonstrates the feasibility of classifying tau burden in amyloid-positive cohorts with magnetic resonance imaging- and amyloid positron emission tomography-derived features and may serve as a surrogate biomarker.

Classification of tau status with machine learning models in amyloid-positive cohorts.

Classification of tau status with machine learning models in amyloid-positive cohorts.

Related papers