Prognostic performance across Alzheimer’s biomarkers, multi-modal physiological measures, and clinical history in asymptomatic individuals
Bibliographic record
Abstract
Abstract Importance Evaluating prognostic performance of Alzheimer’s biomarkers, multi-modal physiological measures, and clinical history in asymptomatic individuals versus established risk factors in asymptomatic individuals is can inform effcient screening strategies. Objective To determine and compare the prognostic performance of amyloid biomarkers, multi-modal physiological measures, and clinical/modifiable risk fac-tors 1 , we conducted a modality-wide assessment of predictors of AD (MODAL-AD) in cognitively asymptomatic patients. Design We used clinical trials (A4/LEARN), longitudinal cohorts (ADNI, AIBL, HABS, NACC, OASIS), and the UK Biobank spanning 2004-2025 (median follow-up time range: 1.8-13.72 years) in time-varying survival and binary classification analyses. Setting Settings included a United States clinical trial, longitudinal cohort studies spread across medical centers in the United States and Australia, and the volunteer-based UK Biobank. Participants Patients were cognitively asymptomatic and age 65+ at baseline, and potentially progressed to either clinical impairment, clinical AD diagnosis, or incurred AD ICD-codes. Patients were volunteer or convenience samples. Exposures PTau-217, amyloid-PET, CSF markers (AB1-42, pTau-181, total-Tau), plasma proteomics, multimodal brain-imaging, and cognitive tests were evaluated as predictors, along with demographics (age, sex, education), APOE geno-type, and modifiable risk factors in the 2024 Lancet report 1 . Main Outcome(s) and Measure(s) PTau-217 and amyloid-PET from A4/LEARN were used to predict clinical impairment (CDR score of 0.5+ on two consecutive visits). PTau-217, amyloid-PET imaging across five cohorts, and CSF markers were used to predict clinical AD diagnosis. Plasma proteomics, multimodal neuroimaging, and cognitive assessments from the UK Biobank were used to predict AD ICD-codes. Results Sample-sizes ranged from 356-28,533 (31-519 cases; female percentages: 48.45-67.39). Models of demographics, APOE genotype, and risk-factors as predic-tors did not show statistically significant differences in time-dependent area under the receiver operating characteristic curve (AUROC) compared to separate models using amyloid biomarkers. Predicting cognitive impairment in A4/LEARN, pTau-217 improved AUROC by 0.045–0.084 (best: 0.616 (CI: 0.51-0.723) vs. 0.7 (CI: 0.609-0.793)). Amyloid-PET improved AD prediction (maximum AUROC increase 0.074; 0.561 (CI: 0.468-0.653) vs. 0.635 (CI: 0.537-0.733)), and CSF biomarkers showed slightly larger gains (maximum AUROC increase 0.127; 0.627 (CI: 0.438-0.816) vs. 0.754 (CI: 0.577-0.931)). In UK Biobank analyses, mean AUROC improvements were minor across proteomics (0.044), neuroimaging (0.143, with 99.8%/0.2% class-balance), and cognitive tests (0.064). Conclusions and Relevance In cognitively asymptomatic populations, biomarkers offer limited advantage over demographics, APOE genotype, and modifi-able risk factors, supporting their importance in early AD screening strategies. Key Points Question How does the prognostic performance of amyloid biomarkers (i.e., pTau-217, amyloid-PET, cerebrospinal fluid markers) and discovery-driven modalities (i.e., plasma protoemics, multimodal brain imaging, cognitive tests) compare to demo-graphics and modifiable risk factors for predicting clinical impairment, clinical AD diagnosis, and AD ICD code outcomes in asymptomatic patients? Findings In this prognostic study of > 300,000 patients, across cohorts, physiological modalities, and outcomes, predictive performance of demographics and modifiable risk factors did not statistically significantly differ from amyloid biomarkers, plasma proteomics, and other modalities. Meaning Alzheimer’s screening in asymptomatic patients can benefit from incorpo-rating modifiable risk factors as additional predictors to amyloid biomarkers.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.007 | 0.012 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".