🤔 How Participants Move from Level 1 to Level 2 in PREDICTOM?

Level 1: Population Scale Risk Stratification

From the published PREDICTOM study methodology article, our Level 1 assessment relies on variables routinely available in primary care or via structured participant input, including demographics, education, medical history, cardiometabolic risk factors, lifestyle indicators, social factors, and family history of dementia. These inputs reflect established dementia risk determinants as synthesised by the Lancet Commission and are feasible for real world implementation.

PREDICTOM researchers estimate risk using validated machine learning time to event models with competing risk adjustment, trained on large longitudinal cohorts such as UK Biobank. Comparative benchmarking against traditional baselines (age only and age sex models) shows a clear performance gain: multivariable approaches achieve mean AUC values of approximately 0.80, with stable concordance, and can predict incident dementia 10–15 years before clinical diagnosis. See Figure 1. This predictive horizon supports strategic upstream intervention planning rather than late stage case identification.

Importantly, Level 1 outputs are probabilistic risk estimates, not diagnostic determinations, ensuring appropriateness for screening and population health use.

Risk Enrichment and Transition to Level 2

Participants are distributed into three different risk groups, lower, intermediate and higher risk, according to pre-defined risk thresholds. Those within higher risk strata are invited to progress to Level 2, forming an enriched cohort with an increased prior probability of early pathological or digital disease signatures. This approach, reflected in the study design schematic, maximises information yield while minimising unnecessary downstream testing.

From a system perspective, risk enrichment improves clinical efficiency, reduces false positive referrals, and aligns intensity of assessment with expected diagnostic value. We also include lower-risk group in Level 2 for comparison and validation.

Level 2: Deep Phenotyping and Biomarker Validation

Level 2 focuses on deep phenotyping, including advanced digital cognitive assessments and biological sampling (for example, blood based biomarkers), depending on site capabilities. Concentrating these resource intensive procedures within a risk enriched cohort enhances statistical power, biomarker signal detection, and validation efficiency, accelerating translational readiness.

Value Proposition for Health Systems and Funders

This two level design delivers a strong impact: low cost, scalable risk screening at Level 1 coupled with targeted high value assessments at Level 2. For health systems, it offers a credible pathway to earlier, risk informed referral and prevention strategies. For policymakers, it represents a scalable, implementation ready framework linking AI based population screening to clinically and biologically meaningful outcomes.

Taken together, the PREDICTOM Level 1 to Level 2 progression illustrates how risk stratified design can bridge population health, clinical research, and future routine care—supporting earlier detection of Alzheimer’s disease with measurable efficiency gains.

Why Your Participation Matters

Every participant plays a vital role. By taking part—even at Level 1—you contribute to a new, smarter way of identifying Alzheimer’s disease earlier, using tools that could one day be part of routine care across Europe.

Participation today helps ensure better prevention, fairer access, and earlier support for tomorrow. Whether or not you move to Level 2, your involvement advances research that aims to change how dementia is detected for everyone.

Joining PREDICTOM study is a simple but powerful step toward a future where Alzheimer’s disease is found earlier—and faced together.