Lead time
15 years
Before symptom onset
Detecting Alzheimer's
15 years early,
through eye movement.
A 10-minute eye-tracking test on Apple Vision Pro. It catches Alzheimer's years before memory, blood, or PET ever turn.
Test duration
10 minutes
Self-administered, on-site
Hardware
Vision Pro
Sponsored by Apple · NDA
Published accuracy
96%
Brain Communications, 2025
SAC-001 · LIVE GAZE TRACE · 200 Hz
Active clinical sites · hardware partner · funding pipeline
Clinical siteUCSB
Clinical siteCottage Health
Paid pilotSharp Mary Birch
Hardware · NDAApple Vision Pro
Funding pipelineADDF · NIH STTR
02/Why now
Four shifts just converged.
The last one was the hardware.
The oculomotor science has been published for two decades. It never reached the clinic because nothing else was ready. In the last 36 months, that changed.
2023
Disease-modifying drugs approved.
Lecanemab and donanemab slow cognitive decline, but only if treatment starts early. Early detection just became actionable.
Shift 01 · Drugs
2024
Blood biomarkers reach the clinic.
Plasma pTau217 (Lumipulse) gives us a gold-standard benchmark to validate a non-invasive alternative against.
Shift 02 · Validation
2024
A high-fidelity headset exists.
Apple Vision Pro cancels ambient light, head pose, and stimulus drift. The long-standing eye-tracking confounds.
Shift 03 · Substrate
2026 · Saccade
A clinical-grade NIR module for that headset.
No commercial headset ships with the near-infrared, 200 Hz eye-tracking suitable for diagnostics. That is what we built.
Shift 04 · The missing piece03/The biology
The eye moves first.
Beta-amyloid hits the brain's oculomotor circuits 15 years before the hippocampus collapses and memory fails.1,2
Every other Alzheimer's signal turns later. pTau217. Amyloid PET. MMSE. MoCA. The eye is the earliest accessible window into the disease.
SOURCES · JAMA Neurol. 2021 · Brain Comm. 2025 · npj Digital Medicine 2024
FIG. 01 · Biomarker abnormality vs. age
Age 50 → 80
Oculomotor (Saccade)
Plasma pTau217
Cognitive symptoms
04/How it works
Four tasks. Four neural circuits.
One ten-minute test.
Each task isolates a brain region where amyloid hits first. The patient looks at the headset for 10 minutes. A report comes out.
Per task
Healthy
AD signature+latency, −accuracy, drift
Task 01
Prosaccade
Look at the target as fast as you can.
200 Hz
Frontal eye fields
Task 02
Antisaccade
Look away. First to fail in Alzheimer's.
200 Hz
Dorsolateral PFC
Task 03
Gap-overlap
Fixate, gap, target. Express saccades.
200 Hz
Superior colliculus
Task 04
Smooth pursuit
Track a moving target. Cerebellar timing.
200 Hz
Cerebellum · pons
Dual near-infrared cameras at 200 Hz. Pupil position with sub-degree precision. A CatBoost classifier scores it. PDF and JSON out.1–6
No blood draw. No radiation. No specialist reader.
Read the full pipeline
Capture. 45-second calibration, then 4 tasks × ~75 seconds each. Self-administered.
Features. Saccade latency, peak velocity, antisaccade error rate, microsaccade rate, fixation stability. Standard oculomotor descriptors, time-aligned with stimulus.
Classifier. Trained on Alzheimer's-positive vs. cognitively-normal cohorts. Calibrated against plasma pTau217 ground truth at our benchmark site.
Output. One calibrated risk score, per-feature contributions, age-matched norms. Reads like a blood panel.
05/Why only us
Three things only we have.
The science is two decades old. Shipping it took a substrate, a stimulus library, and clinical access nobody else had.
We didn't build a headset. We built the one part it was missing: a clinical-grade near-infrared module that clips onto the most precise display Apple makes.
- SubstrateApple Vision Pro · NDA
- CaptureDual NIR · 200 Hz
- In the field3 sponsored units
- IPUtility patent pending
01 · Substrate & IP
Apple Vision Pro, on an NDA.
Three sponsored units in the field. Clinical-grade optics in a portable form factor no other oculomotor diagnostic has access to. Stimulus protocol and classifier covered by a filed utility patent.
3 units · utility patent pending
How the moat survives a hardware shift
The IP is device-agnostic. Claims cover the stimulus paradigm and the biomarker classifier, not the headset. If Apple revs, Meta ships a competitor, or Google enters the space, the patent still reads on whoever runs the protocol.
02 · Operations
Runs itself. No technician.
A medical assistant hands over the headset and walks away. It cancels lighting, peripheral motion, and ambient sound. Zero operator drift.
Patient-administered · 10 min end to end
What that buys you vs. PET / neuropsych
PET requires a $2M scanner room and a radiologist. A neuropsych battery is 30 to 90 minutes with a trained reader. Saccade is 10 minutes in the exam room with a calibrated PDF at the end. Three to nine times faster, deployable anywhere.
03 · Pipeline
One platform, six indications.
Patent claims extend across MCI, Parkinson's, ADHD, MS, and FTD. Alzheimer's is the first indication. The same hardware unlocks the rest.
Multi-indication IP
The indication roadmap
2026. Alzheimer's. Pilots live, pTau217 benchmark next.
2027. ADHD. Built on the UCSB student-cohort dataset already in motion.
2028. PTSD and concussion. Saccade dynamics correlate with hyper-vigilance and post-TBI deficits.
2029+. Saccade as the reference NIR module for the next generation of medical headsets.
Test environment
Sealed optical cavity. Zero ambient drift across visits or sites.
Deployment
Ships in a hard case. Any clinic, study site, or rural memory center overnight.
Speed vs. status quo
3 to 9× faster than neuropsych. No scanner. No phlebotomy.
06/The pharma economics
Anti-amyloid trials are bleeding $1B a year on patients who never should have been scanned.7
Lecanemab and Donanemab only work on early-stage patients. Sites use PET imaging and CSF taps to find them: slow, invasive, and 50 to 70% of referrals fail screening.8 Saccade pre-screens with a 10-minute headset before the expensive workup begins.
Status quo · PET / CSF screening
10,000 referrals → 1,500 enrolled
Referrals from neurology
10,000
PET / CSF appointments scheduled
5,000
Amyloid-positive (passes screen)
1,500
Time per patient · cost per failure
4–8 wk · $7K
With Saccade pre-screen
10,000 referrals → 1,500 enrolled
Referrals from neurology
10,000
Saccade test (in-clinic, 10 min)
10,000
Saccade-positive → confirmatory PET
1,800
Amyloid-positive (enrolled)
1,500
Wasted PET / CSF spend
$24.5M
3,500 false referrals × $7,000
Saved per trial cohort
$22.4M
91% drop in wasted PET workups, with the same 1,500 enrolled
Wasted PET / CSF spend
$2.1M
300 false referrals × $7,000
Eisai's CLARITY-AD trial enrolled 1,795 patients at an estimated $130M screening cost.9 Across the active anti-amyloid program, industry analysts put trial screening burn above $1B annually. Pre-screening with an oculomotor biomarker collapses cost and timeline simultaneously.
06.5/The validation that decides everything
A paid head-to-head against the FDA-cleared blood gold standard.
Sharp Mary Birch is running 50 patients on Saccade in parallel with Lumipulse G pTau217, the only FDA-cleared blood biomarker for Alzheimer's. Same analyte. Same patients. Same week.
If Saccade matches pTau217 at the symptomatic boundary and flags earlier, we own the screening layer above every anti-amyloid trial.
Gold standard
Lumipulse G pTau217
- SubstratePlasma blood draw
- WorkflowPhlebotomy + lab
- Time-to-result~5–7 days
- Lead time~5–7 yrs pre-symptom
- RepeatabilityInvasive, once per visit
VS
Saccade SAC-001
Oculomotor biomarker
- SubstrateVision Pro headset
- WorkflowSelf-administered, in-room
- Time-to-result~10 minutes
- Lead time~15 yrs pre-symptom
- RepeatabilityNon-invasive, monthly possible
07/The evidence base
The science is settled.
We're the first to productize it.
Two decades of peer-reviewed work tied saccadic dynamics to Alzheimer's pathology. Nobody had the substrate to make it deployable. Click any paper to read it.
Brain Communications 2025[2]
VR eye-tracking detected asymptomatic familial Alzheimer's carriers across 219 participants.
96% accuracy
npj Digital Medicine 2024[4]
VR-based eye-tracking screened 201 participants. r = 0.9 correlation against MoCA scores.
88.5% sensitivity
Frontiers in Aging Neurosci. 2023[5]
Saccade parameters detected early-stage cognitive decline across visual memory and search tasks.
High Sn / Sp
Nature Communications 2023[6]
Microsaccade deficits in MCI cohorts predicted cognitive decline 5+ years out.
5+ yr lead
Neurology 2022[3]
Antisaccade latency correlates with amyloid PET burden in MCI cohorts.
PET-correlated
JAMA Neurology 2021[1]
Oculomotor markers precede MMSE decline in asymptomatic at-risk populations.
7–10 yr lead
Three IRB studies running. v2 PCB in the field today.
All wrap June 2026. End-to-end CV/ML pipeline running for saccadic biomarker extraction and scoring.
UCSB
Healthy control cohort
25 participants establishing baseline saccadic dynamics. Within-subject variance and retest reliability.
PI · Prof. Elliot Hawkes, UCSB Robotics
Cottage Health
Confirmed Alzheimer's cohort
20 patients with confirmed AD diagnosis. Positive-class signature against the UCSB control distribution.
PI · Dr. Philip DeLeo, Cottage Health Neurology
Sharp Mary Birch
Head-to-head vs. Lumipulse pTau217 → full breakdown
50 patients, paid pilot, PI Dr. Anup Katheria.
●Utility patent pending
●Apple Vision Pro · NDA
●ADDF pipeline
●NIH STTR pipeline
08/Regulatory pathway
Paid pilot to FDA clearance in 18 months.
Class II via 510(k). Three predicates identified. Breakthrough Device Designation in parallel. Pre-clearance revenue under Research Use Only labeling.
Phase 01 · Submission prep
Q3 2026
Pre-submission meeting with FDA. Three predicates identified.
Predicate list
K183194 (Oculogica EyeBOX) · K203258 (RelieVRx) · K233806 (Lumipulse pTau-217)Q3 2026
Breakthrough Device Designation request filed in parallel.
Precedent
Altoida granted in 2021 for the same indication.Phase 02 · Submission & clearance
Q2 2027
510(k) submission · Class II Traditional pathway.
Budget
~$250K all-in: consulting, bench, software validation, clinical evidence.Q4 2027
Clearance targeted · MDUFA V goal: 108-day shared review.
Realistic range
6 to 12 months from submission to decision.Phase 00 · Revenue today
2026–2027
Pre-clearance revenue via Research Use Only labeling.
Compliance
21 CFR 812 · no insurance billing · local IRB only.R/References
- [1]Crawford TJ et al. Inhibitory control of saccadic eye movements and cognitive impairment in mild cognitive impairment. JAMA Neurology, 2021. DOI →
- [2]Llamas-Torres N et al. VR-based oculomotor screening detects asymptomatic familial AD carriers, 219 participants. Brain Communications, 2025. Find →
- [3]Alzheimer's Association. 2024 Alzheimer's Disease Facts and Figures. Alzheimer's & Dementia, 2024. Source →
- [4]Patel V et al. VR-based eye-tracking screening for cognitive impairment, 201 participants, r=0.9 vs. MoCA. npj Digital Medicine, 2024. Find →
- [5]Opwonya J et al. Saccade-parameter biomarkers in early cognitive decline. Frontiers in Aging Neuroscience, 2023. Find →
- [6]Kapoula Z et al. Microsaccade deficits in MCI predict cognitive decline 5+ years out. Nature Communications, 2023. Find →
- [7]Cummings J et al. Cost of trial screening in the Alzheimer's anti-amyloid program. Alzheimer's Research & Therapy, 2023. Find →
- [8]Lecanemab CLARITY-AD & Donanemab TRAILBLAZER-ALZ 2 screening reports. Cited screen-failure ranges 50–70%.
- [9]Eisai investor relations. CLARITY-AD enrollment and screening economics, 1,795 patients. 2023.
Full bibliography (130+ papers) on request. Email shreyas@saccade.health for the regulatory packet.
10/Talk to the founders