# NDx — Precision Neurology

> AI-enabled multi-omic blood tests for neurodegeneration. We help pharma enroll the right patients and bring successful therapies to patients.

Text version of the NDx pre-pitch deck (Sept 2026): one section per slide, in slide order, with the same wording and figures as the slides. Figures marked modeled, target or illustrative are NDx models or plans, not measured results.

## The Problem — The measurement gap in neurodegenerative diseases

The gap hampers drug development, clinical monitoring and, ultimately, patient care.

- **Diagnosis lags behind.** >50M people live with dementia ([WHO](https://www.who.int/news-room/fact-sheets/detail/dementia)); 3 in 4 cases are never diagnosed ([ADI](https://www.alzint.org/about/dementia-facts-figures/dementia-statistics/)). Global cost of dementia: $1.3T a year ([WHO](https://www.who.int/news-room/fact-sheets/detail/dementia)).
- **Current tests are expensive, invasive or limited.** ~$5,000 per PET scan ([USC](https://schaeffer.usc.edu/research/key-barriers-for-clinical-trials-for-alzheimers-disease/)), or an invasive spinal tap. Blood tests read Alzheimer's only ([FDA](https://www.fda.gov/news-events/press-announcements/fda-clears-first-blood-test-used-diagnosing-alzheimers-disease)).
- **Pharma suffers in drug development.** ~70% of trial candidates screen-fail, then it takes 12–18 months to see if the drug works ([Alz&Dem](https://alz-journals.onlinelibrary.wiley.com/doi/10.1002/alz.71681)).

## The Solution — Bridging the measurement gap with AI-powered multi-omic blood tests

1. **Biological foundation:** in neurodegeneration, dying brain cells shed molecular signals into the plasma.
2. **Multi-omic data:** two signals from the blood.
   - cfDNA 5mC: reflects real-time molecular signals and disease state.
   - Autoantibody profile: profiles personalized immune responses over the longer term.
3. **AI fusion:** AI-enabled multi-omic integration combines these signals with clinical and biological knowledge to identify biomarkers.
4. **Niche market, pharma clinical trials:** the initial focus is molecular patient stratification; future expansion is progression and treatment-response monitoring.

Built for scalable, AI-driven biomarker development.

## AI Roadmap — From research assistant to autonomous AI scientist

An AI-native path to blood diagnostics for neurology.

| Phase | Status | How it works | Key focus | Deliverable |
|---|---|---|---|---|
| 1. AI Research Assistant | In development | Scientists guide the agent step by step | Data engine; pipeline automation | Agentic biomarker analysis and development pipeline |
| 2. AI Research Collaborator | Planned | Contributes its own judgement alongside the team | Benchmarking & evaluation; classical ML models | Biomarker benchmark + classical ML models |
| 3. Autonomous AI Scientist | Long-term vision | Trustworthy judgements, scientists sign off | Recursive self-improvement; blood foundation model | Recursive self-improvement loop |

## Our Current State — Human proof-of-concept completed, ready to scale up

- **50 patients** completed a 2-year pilot.
- **3 diseases analysed:** AD, PD, CBS.
- **15 customer conversations** with 9 companies, 4 of them big pharma; 4 more scheduled.
- **$200K angel funding** offer received.

Done so far:

- Multi-omic pilot datasets generated
- 2 orthogonal assay platforms
- Core multi-omic workflows established
- AI-assisted biomarker discovery pipeline established
- Stanford biobank collaborations in place
- Stanford OTL invention disclosure filed

## Next Stage — Building the data foundation for AI-driven diagnostics

Scale high-quality paired measurements while establishing clinically relevant biomarker value.

- Month 6 target: ~320 initial paired plasma samples.
- Month 18 target: 2,000+ total paired plasma samples.
- Per plasma sample: 2 assays, cfDNA 5mC + autoantibody profiling.
- Initial disease coverage: Alzheimer's, Parkinson's, ALS and healthy controls.

Three validation objectives:

1. **Multimodal value:** complementary biology, patient heterogeneity and use-case utility.
2. **Longitudinal relevance:** biomarker trajectories linked to clinical change; reproducibility & QC.
3. **Translational readiness:** candidate panel, locked AI model and pilot-ready assay / report.

Target: longitudinally informative clinical annotations; final cohort mix and repeat-sampling allocation to be defined with partners.

## Go to Pharma — A screening, monitoring platform for pharma (illustrative use case)

**1. Pre-screening: who should enter the trial?** An accessible, rapid, affordable blood test to screen patients for clinical trials.

- The NDx blood screen runs on all 1,000 candidates and ~300 go on to PET, so PET only confirms the shortlist.
- PET today: ~$5,000/scan. NDx, modeled: ~$950/sample.
- PET spend drops from 1,000 scans to ~300. Target turnaround: ~2 weeks.

**2. Longitudinal monitoring: is the drug working?** Objective, blood-based molecular monitoring of treatment effect.

- PET + CSF: ~18 months. NDx target: 6–12 weeks.
- ~15 months: modeled earlier trial decision.
- ~60%: modeled reduction in test costs.
- Goal: an earlier molecular treatment-effect signal, for faster go / no-go.

Illustrative scenario: per-sample pricing, turnaround and savings are NDx models on stated assumptions, not measured product performance.

## Market — A big, open market with high growth

| | Per year | Definition |
|---|---|---|
| TAM | $12.8B | All brain-disease diagnosis and monitoring |
| SAM | $5.05B | Neurodegenerative disease specifically: AD, PD, ALS and related |
| SOM | $1.65B | Our 5-year goal as the category leader, ~30% of SAM |

- **The category just opened:** the FDA cleared the first Alzheimer's blood test in May 2025; Labcorp launched it nationwide that August ([FDA](https://www.fda.gov/news-events/press-announcements/fda-clears-first-blood-test-used-diagnosing-alzheimers-disease), [FMI](https://www.futuremarketinsights.com/reports/blood-based-biomarker-for-alzheimers-disease-diagnostics-market)).
- **And it compounds:** blood-based AD biomarkers go from $169M (2025) to $530M (2033), a 3.1× move ([FMI](https://www.futuremarketinsights.com/reports/blood-based-biomarker-for-alzheimers-disease-diagnostics-market)).
- **Capital is following:** Gates is funding blood-based Alzheimer's diagnostics and calling for routine screening from age 60 ([Gates Notes](https://www.gatesnotes.com/work/cure-alzheimer-s)).

TAM is the global annual brain-disease diagnosis and monitoring market; SAM is the neurodegenerative-disease segment; the ~30% 5-year share is an NDx assumption. Sources: growth [FMI](https://www.futuremarketinsights.com/reports/blood-based-biomarker-for-alzheimers-disease-diagnostics-market); clearance [FDA](https://www.fda.gov/news-events/press-announcements/fda-clears-first-blood-test-used-diagnosing-alzheimers-disease), May 2025; demand signal [Fortune](https://fortune.com/2025/11/10/bill-gates-alzheimers-blood-tests-routine-medicals/), [STAT](https://www.statnews.com/2025/06/17/bill-gates-alzheimers-biomarkers-blood-based-diagnostic-tests-medication/).

## Business Model — Pharma pays today, the data compounds

An AI-native biomarker engine where commercial adoption, proprietary data and validation reinforce each other.

The NDx flywheel: pharma contracts (revenue + clinical problems) → paired data (longitudinal multi-omics) → AI learning (validated insights) → better products (more pharma adoption) → back to pharma contracts.

- **Who pays:** pharma & biotech sponsors, through study contracts + per-sample analytical fees.
- **How we scale:** trials → programs → indications, with reusable assays, AI models and sponsor relationships.
- **What compounds:** proprietary data + validated pan-ND AI, strengthened by pharma partnerships and repeated learning.

Go-to-market, land & expand:

1. Retrospective studies: paid studies on existing trial samples.
2. Prospective trials: longitudinal monitoring & validation.
3. Portfolio expansion: more programs, sponsors & indications.

Longer-term opportunity: companion diagnostics and clinical monitoring.

## Competition — Biology-guided AI, built around pharma needs

Differentiation today: signal selection + customer focus. Defensibility tomorrow: data + validation + workflow integration.

Strategic positioning, illustrative and not a performance ranking. One axis is customer focus (technology / model-led to pharma customer-led), the other biological measurement strategy (broad data integration to biology-guided signal selection):

- Pharma customer-led, biology-guided signal selection: NDx.
- Pharma customer-led, broad data integration: NeuroDex, Circular Genomics, Valinor.
- Technology / model-led, broad data integration: Network Bio, Prima Mente.
- Technology / model-led, biology-guided signal selection: Durin.
- Test modalities (PET imaging, CSF biomarkers, emerging blood tests) sit toward broad data integration.

The moat we build:

1. **Biological insight:** purposeful pairing of cfDNA methylation and autoantibody signals.
2. **Proprietary learning data:** matched longitudinal profiles connected to disease trajectories.
3. **Pharma integration:** evidence for stratification, progression and trial workflows.

Better data → better models → stronger clinical evidence. Hypothesis today; a validated, compounding advantage over time.

## Team — Built to take biomarkers from discovery to clinics

Wet lab, computational biology and ML product, with the clinical access to run the study.

- **Ze Yang**, Co-founder & CEO: molecular biology & multi-omics. Scientist, Stanford University; PhD, Stanford University.
- **Daniel Zheng**, Co-founder & CTO: AI/ML & platform engineering. Siri AI, Apple; former CTO, Utopia Studios.
- **Siyu He**, Scientific Co-founder & CSO: computational biology & biomarker discovery. Postdoc, Stanford University; PhD, Columbia University.
- **Kathleen Poston**, Clinical Advisor: clinical neurology & biomarkers. Professor of Neurology, Stanford Medicine.

## Contact

NDx: building accessible AI for brain health. [yangze@stanford.edu](mailto:yangze@stanford.edu) · Sept 2026