Wireframe anatomical render of a human brain, representing NeuroFusion-AD's neurology focus
NFAD · Neurology — Alzheimer's DiseaseRetrospective research

NeuroFusion-AD

A multimodal research model estimating amyloid-positivity and cognitive-decline risk in patients with mild cognitive impairment.

• Work with us
One
•  The architecture

A graph neural network that fuses four clinical streams into a single patient-level prediction.

NFAD — Neurology — Alzheimer's Disease

Each patient is modelled as a heterogeneous graph — biomarker, speech, gait and demographic nodes connected by learned weighted edges — capturing inter-modal correlations that single-modality models miss.

Plasma biomarkers

pTau181 and NfL — two numeric inputs drawn from a routine blood draw.

Speech analysis

A 60-second Cookie Theft picture description, reduced to acoustic and linguistic features.

Gait analysis

30 seconds of smartphone accelerometry, reduced to spatiotemporal gait features.

Demographics

Age, sex, APOE4 genotype and MMSE score.

0.907
Amyloid positivity AUC — Bio-Hermes-001 test set
0.8897
Amyloid positivity AUC — ADNI test set
0.651
Dementia progression C-index

Retrospective research results on held-out research datasets. They are not clinical performance claims, and NFAD is not available for clinical use.

Two•  The clinical problem
[ IMAGE · 16:7 ]
[ IMAGE · 16:7 ]

Built for the years before symptoms

Alzheimer's pathology begins 15–20 years before symptoms, yet 90% of patients are diagnosed at a moderate or late stage. NFAD is being researched as a possible primary-care triage aid, where a higher-risk estimate could support earlier specialist referral.

Why it matters
15–20 years

Pathological damage begins long before the first symptom appears.

90% diagnosed late

Most patients are identified only at a moderate or late stage.

$3,500–$8,000

The cost of confirmatory amyloid PET, inaccessible to most patients.

6-month waits

Average specialist wait times for those seeking assessment.

Three•  How it was built

Four phases of retrospective research

NFAD was developed in sequence, with a mid-project fix to eliminate data leakage before evaluation on two held-out research test sets. Final validation on DementiaBank is on hold pending incorporation.

01
Architecture & pipeline

Graph design and the multimodal data pipeline established.

02
Model development

Initial model trained and tuned on ADNI data.

03
Validation & leakage fix

Data leakage resolved by reducing model capacity and rebuilding the train–test split protocol.

04
Retrospective evaluation

Performance measured on held-out ADNI and Bio-Hermes-001 research test sets. Not a clinical performance claim.