MCI → AD · 24-Month Progression Risk
Predict 24-Month Alzheimer’s Disease Progression
from Routine Clinical Data
Clinovia uses validated machine learning models to estimate progression risk in patients with Mild Cognitive Impairment (MCI). Start with routine cognitive testing, and improve prediction when MRI measurements are available.
For Research Use Only. Not a diagnostic device.
Why Clinovia
Built around how memory clinics already work
Uses data you already collect
No new tests, no new equipment — just the routine cognitive workup most memory clinics already perform.
MRI when available
No workflow changes. Enter MRI measurements if available, and Clinovia automatically applies a more informative model.
A report in seconds
A complete risk assessment, ready before the visit ends.
One Workflow
You never choose the model. Clinovia does.
One intake replaces separate “Cognitive” and “Cognitive + MRI” products. Whatever data you enter, Clinovia routes it to the correspondingly validated model — automatically.
Patient Information
Age, sex
Cognitive Assessment
MMSE required · RAVLT or LIMM
MRI
Optional
Clinovia
Selects the most appropriate validated model
| Data Available | Model Applied | AUC |
|---|---|---|
| MMSE only | Clinical (MMSE) | 0.87 |
| MMSE + RAVLT | Clinical (RAVLT) | 0.91 |
| MMSE + LIMM | Clinical (LIMM) | 0.90 |
| MMSE + RAVLT + LIMM | Clinical (Complete) | 0.92 |
| Any of the above + MRI | Clinical + MRI | 0.91 – 0.93 |
Every combination is routed to a separately validated model — nothing is estimated from a partial or imputed feature set. AUC figures are cross-validated (5×10 repeated CV); see Validation for confidence intervals and methodology.
Inputs
Everything Clinovia needs, and nothing more
Patient
- Age
- Sex
Cognitive
- MMSE
- RAVLT Immediate Recall
- LIMM Total
MRI
Optional- Hippocampus
- Entorhinal
- Mid Temporal
- Whole Brain
- Ventricles
Example Report
A complete answer, not just a score
24-Month Risk
Recommendation
Routine Monitoring
Model Used
Clinical + MMSE + RAVLT
Sample report for illustration. For Research Use Only.
Validation
Cross-validated on ADNI. External validation underway.
Development Cohort
ADNI
n = 2,430
Validation Method
5×10 CV
Bootstrap 95% CI
AUC — Clinical Models
0.87 – 0.92
AUC — Clinical + MRI Models
0.91 – 0.93
External Validation
A note on precision
At each model’s optimal risk threshold, roughly 3 in 5 patients flagged high-risk go on to progress within 24 months. The models are stronger at ruling out low risk (NPV ≈ 0.96) than at confirming high risk — worth keeping in mind when interpreting an individual report.
Robustness across model architectures
Episodic memory measures carrying the dominant predictive signal, with MRI providing the largest benefit when cognitive testing is limited, held consistently across two independent model architectures — logistic regression and gradient boosting.
No hype. Just evidence.
Designed for Memory Clinics
Built for how the data is actually used
Routine Assessment
Fits into an existing MCI workup — no new equipment or tests required.
Retrospective Validation
Apply Clinovia to historical patient data to see how it would have performed.
Clinical Research
Support cohort characterization and enrichment analyses.
Trial Screening
Help identify likely progressors for enrollment in progression-focused trials.
Pilot Program
See it work in your clinic, before you commit to anything more.
FAQ
Questions clinicians ask first
Ready to Explore Clinovia?
We're looking for pilot clinics and research partners to bring first-contact MCI risk stratification into practice.