Where this came from — a personal mission, not a clinical product
This didn't start as a business plan. I live with ADHD and bipolar disorder. Years of personal research into how attention, mood, and sensory load actually work convinced me that local, on-device inference over personal behavioral signal has real research potential — for conditions like these, and others: epilepsy, Alzheimer's, PTSD.
To be direct: this is a research direction, not a clinical product. I have no medical credentials, and nothing here is clinically validated. But the underlying concept holds, funding exists specifically for this kind of independent research, and it's a mission I intend to pursue seriously, with real research partners.
The sensing stack envisioned draws on commercially available devices — the Muse 2 biofeedback headband, EKG nodes concealed in headwear, optical smartwatch sensors for HRV and temperature, and vocal acoustic transducers. Because edge inference keeps raw neurological streams on the local NPU, each person retains complete ownership of their cryptographic keys and their biometric data never touches a cloud server. Edge inference means you carry your own hypervisor in your pocket — your graph, your keys, your device.
Research hypotheses — not present capabilities
Hypothesis: delay discounting curves + motor variance → chunked micro-tasks, deadline guardianship.
Hypothesis: sleep architecture shifts + sentiment trajectories → cycle-transition early warning.
Hypothesis: vagal withdrawal, tremor, kinetic agitation → grounding haptics, paced breathing.
Hypothesis: longitudinal degradation curves, pre-ictal shifts → risk windows for care teams.
None of these are built. They are the research directions this substrate was designed to explore. HIPAA posture would be via hardware-backed local encryption — not yet implemented.
Research matrix — hypotheses, not outcomes yet
Research hypotheses only. No diagnostic claims, no predictive guarantees. What follows is the signal each research direction targets and the design goal it's aimed at — not a validated output. Zero health data leaves the device — a design constraint, not yet a delivered system.
| Condition | Signal hypothesized | Hypothesized scaffolding |
|---|---|---|
| ADHD | Dopamine kinetics, delay discounting, motor variance | Chunked micro-tasks, deadline guardianship |
| Autism (ASD) | Sensory load, acoustic density, HRV | Pre-meltdown soft-muting |
| PTSD / Panic | Vagal withdrawal, tremor, kinetic agitation | Grounding haptics, paced breathing |
| Bipolar | Sleep shifts, sentiment trajectories | Cycle-transition early warning |
| Alzheimer's / Epilepsy | Degradation curves, pre-ictal shifts | Longitudinal record, research-stage risk flagging for care teams |
Concept topology — Biomedical research (60/40 weaver)
{
"status": "NODE_UNSELECTED",
"instruction": "Click any node or execute a vector query above to inspect parameters."
}The same fabric, commercialized — one-way mirror
This is a research direction. But the on-device, local-first substrate it's built on already powers commercial services that could never route sensitive data through a public cloud:
Air-gapped appliance that learns your corpus locally and answers in your language — no cloud account, no outbound pipe, compliance by architecture.
Research discloses. Commercial delivers. Same fabric — one-way mirror.
Where this stands
This is a research direction, not a clinical product. No medical credentials. No clinical validation. No biometric pipeline built. The substrate exists; the research programs need funding and partners. See the full record: /partners