Clinical Modules
All Alzheimer's models trained and validated on ADNI data · For research use only
Neurology — Alzheimer's
Random Forest · XGBoost · ADNIStage 1 — Clinical Screening
LiveEstimates 24-month MCI-to-AD progression risk from routine clinical variables. Routes high-risk patients to Stage 2a and/or 2b.
Inputs
- Age, sex, education
- APOE ε4 count
- MMSE
- ECog (study partner total & memory discrepancy)
- RAVLT immediate & forgetting
Outputs
- Risk class: High Risk Progressor / Low Risk Monitor
- 24-month progression probability
- Model confidence
- Top contributing features
- Next step routing
Stage 2a — Plasma Amyloid Triage
LiveEstimates amyloid burden from blood-based biomarkers. Runs in parallel with Stage 2b and feeds the fusion layer.
Inputs
- Plasma Aβ42/40 ratio
- p-tau217
- GFAP
- NfL
- Age, APOE ε4, education, MMSE (covariates)
Outputs
- Amyloid probability (0–100%)
- Classification: Amyloid Positive / Negative
- Risk band: Low / Intermediate / High
- Recommendation: Order / Consider / Defer PET
- Top contributing features
Stage 2b — MRI Neurodegeneration Gate
LiveConfirms or rules out structural neurodegeneration using MRI volumetrics and longitudinal atrophy trajectories. Runs in parallel with Stage 2a.
Inputs
- Hippocampal, entorhinal, ventricular, whole-brain volumes
- Intracranial volume (ICV)
- Longitudinal slopes: hippocampus, ventricles, whole brain
- Age, APOE ε4 (covariates)
Outputs
- MRI risk probability
- Classification: N+ Positive / N+ Negative
- Risk band: Low / Intermediate / High
- Clinical interpretation
- Next step routing
Fusion Layer
Coming SoonCombines Stage 2a plasma and Stage 2b MRI outputs into a final amyloid burden estimate and PET/CSF referral recommendation.
Inputs
- Stage 2a output
- Stage 2b output
Outputs
- Final amyloid estimate
- PET/CSF referral recommendation