USC researchers published a new blood-based algorithm that reduces the need for PET scans in Alzheimer's clinical trial screening from more than 70% to 31%.
On September 9, 2026, researchers at the Keck School of Medicine of USC published a new blood-based screening algorithm in Alzheimer’s & Dementia: The Journal of the Alzheimer’s Association. This statistical model sharply reduces the number of PET scans needed to identify people at risk of Alzheimer’s disease for clinical trials.
The USC researchers developed this screening algorithm to make recruitment for preclinical Alzheimer’s clinical trials significantly more efficient. They specifically designed the model for the AHEAD 3-45 prevention trial. This international Phase 3 trial studies whether earlier treatment with lecanemab could improve outcomes in people with amyloid-beta accumulation but no obvious cognitive symptoms. Lecanemab is an approved Alzheimer’s drug that removes amyloid-beta aggregates and slows clinical progression by approximately 30% in symptomatic patients.
AHEAD-style prevention trials face a major screening challenge because amyloid buildup can begin years or decades before noticeable memory problems appear. Only about 30% of cognitively healthy adults older than 65 were estimated to have amyloid levels high enough to qualify for trials such as AHEAD. Before the blood-based screening approach was introduced, more than 70% of people who underwent PET imaging during recruitment were ultimately found ineligible. PET is a costly imaging method widely used to detect amyloid pathology.
The study treated PET imaging as the diagnostic gold standard. The USC algorithm reduced the proportion of scanned candidates who did not qualify from more than 70% to 31%. The researchers refined the algorithm through three successive versions during active recruitment from 2020 through 2024. A first version of the algorithm was introduced in February 2022.
It used an amyloid-beta ratio and reduced the PET-ineligible proportion from 71% to 50%. A second version arrived in May 2023. This update added the blood marker p-tau217 and reduced the rate further to 31%. The final model utilized these blood-plasma biomarkers alongside age and APOE4 carrier status.
APOE4 is a genetic factor associated with an increased risk of amyloid accumulation in the brain. The study describes the amyloid-beta ratio as an earlier signal of amyloid accumulation, while p-tau217 is a marker that more reliably reflects amyloid burden. The USC researchers built the model using clinical data from 1,080 AHEAD participants. They subsequently validated the algorithm against an independent dataset from the Wisconsin Registry for Alzheimer’s Prevention.
The team utilized a statistical approach called a Mixture of Experts rather than a simple positive or negative result. This method estimated where a participant fell along a continuous spectrum of amyloid accumulation. The researchers designed this approach to identify an intermediate range where blood markers alone may be less decisive.
Clinical trial screening is typically among the most costly parts of a trial for research sites. Oliver Langford is the study’s corresponding author and a simulation director at USC. He stated that the algorithm reduced the burden on sites and patients by reducing the number of people who needed PET scans. The immediate demonstrated benefit is operational efficiency for large-scale research.
For older adults noticing persistent memory or thinking changes, this research highlights a shift in clinical evaluation pathways. A blood test may increasingly become an initial discussion point rather than a final diagnostic answer. Adults over 60 can ask a clinician if an evidence-based blood test is appropriate for their specific situation. This approach can often guide discussions before pursuing costly imaging as a first step.
Getting a clear medical evaluation helps older adults find targeted resources for cognitive protection. Recent research focuses heavily on plasma p-tau217 and amyloid-beta 42/40 measurements. These markers provide biological information related to amyloid pathology through a simple blood draw. A separate 2026 report on plasma p-tau217 described a two-step workflow for preclinical trials.
In that study, a stand-alone p-tau217 test produced a positive predictive value of 79% and an overall accuracy of 81%. When researchers used a two-step workflow, both measures increased to 91%. The same report indicated the workflow reduced confirmatory PET scans by more than 75%. The healthcare sector is also moving toward clinical blood testing for people who already exhibit cognitive symptoms.
Coverage of the FDA-cleared PrecivityAD2 test described it as a tool for adults undergoing evaluation because of memory loss or impaired thinking. It is not intended as a general screening test for people without symptoms. This helps older adults seeking practical guidance on memory understand when a test is actually useful. A blood-based biomarker test may assist a medical evaluation, but it is not an unsupervised checkup for everyone.
Patients and clinicians must understand that blood-based biomarkers are not a standalone diagnostic test for routine screening of the general public. The USC algorithm was explicitly developed for clinical trial recruitment. The study article specifically does not claim that the algorithm replaces PET scans or independently diagnoses Alzheimer’s disease. The reported 31% figure only describes the percentage of people who received PET scans but were found ineligible for the AHEAD trial.
It does not mean that 31% of all older adults have Alzheimer’s disease. It also does not mean that 69% of people tested would receive a definitive diagnosis without imaging. Langford stated that population measurements show an intermediate region of amyloid accumulation. A plasma marker alone cannot fully capture a person's status in this borderline range.
Because the model identifies this intermediate range, a borderline result may lead to additional testing rather than immediate reassurance. The researchers designed the Mixture of Experts approach to model that uncertainty more effectively. Consequently, a blood result may provide a probability or risk estimate rather than a definitive diagnosis in every single case. People with discordant results may still need PET imaging or cerebrospinal-fluid testing.
Furthermore, a 2026 review highlighted that blood test performance varies widely across tests and populations. Across 49 observational studies covering 31 blood-based tests, pooled sensitivity ranged from 49.3% to 91.4%. Pooled specificity ranged from 61.5% to 96.7%. This variation supports the need to consider the specific assay, laboratory method and clinical purpose.
A research participant's eligibility threshold may not be identical to the threshold a clinician would use when evaluating an individual with memory concerns. Consumers should consult their physicians to discuss the specific test being used and the available brain health resources before making medical decisions.
The USC research demonstrates a significant step forward in making clinical memory evaluation less invasive and more accessible. By combining blood biomarkers with age and genetic risk factors, researchers can more efficiently identify candidates for early intervention trials. This development spares many patients the cost and stress of unnecessary brain imaging during the screening process. Less burdensome testing pathways could eventually allow more people to obtain an earlier assessment of Alzheimer’s pathology.
However, access to initial blood testing is not the same as absolute diagnostic certainty. Blood biomarkers indicate biological features associated with Alzheimer’s disease, but they do not explain every cause of cognitive change. Other potentially relevant causes can include medication effects, sleep disorders and depression. Hearing loss and vascular disease are also known to affect cognitive function.
A clinical professional must still interpret any blood result within the context of a patient's specific symptoms, medical history and physical examination. While more accessible testing provides valuable data, these tools serve best as aids to a comprehensive evaluation. For adults over 60, this algorithmic progress offers a less burdensome starting point for medical conversations. It provides a data-driven foundation for older adults to manage their cognitive health proactively alongside their physicians.
Understanding how these tools function helps patients focus on evidence-based approaches to healthy brain aging without relying on inaccurate assumptions.
Patients and their families own the responsibility of discussing memory concerns with their medical providers, and FitBrainLab supports these critical conversations with objective clinical analysis. Difficulty separating established evidence from early or exaggerated health claims creates unnecessary confusion, but FitBrainLab translates complex brain aging research into clear guidance so older adults can evaluate their clinical options calmly. Explore Resources
Follow FitBrainLab for research-led insights on memory, focus, brain aging, nutrition and mental fitness after 60. Stay connected for new articles, practical guidance and ideas for a sharper, more engaged life.
Build habits that support memory, focus and a curious, connected life.
Read the Blog
