ANM
ThesisFrameworkOpportunitiesLibraryField Notes
한국어Admin
AI-Native Medicine
Opportunity Letter한국어
All issues

ISSUE 001 · Oct 3, 2026 · 7 min read

What if stimulation arrived only when a BCI detected motor intent?

Instead of selling another device, design a closed-loop rehabilitation service that learns each patient’s response and adapts timing and intensity.
The issue in one sentence

Instead of selling another device, design a closed-loop rehabilitation service that learns each patient’s response and adapts timing and intensity.

Diagram 01

A closed loop that learns

01Motor intent→
02EEG sensing→
03Confidence gate→
04Adaptive policy→
05FES · robot→
06Observed movement→
07Functional outcome

Reconstructed by ANM from the detection-to-feedback structure shared across BCI-FES trials, with adaptive-policy and outcome-learning stages added.

01Clinical problem · Evidence

Rehabilitation misses the exact moment a patient intends to move.

Many people retain substantial arm impairment after stroke. Training dose and quality vary, while aligning electrical stimulation with true motor intent remains difficult.

02Technology stack · Proposal

Read neural intent and return it as movement.

A non-invasive EEG BCI detects motor intent. AI tracks signal quality and fatigue, then triggers functional electrical stimulation or robotic assistance in real time.

03Research question · Hypothesis

Does patient-level adaptation beat a fixed protocol?

Test whether adapting thresholds, intensity, and tasks to signal reliability and fatigue improves upper-limb function and daily activity at 12 weeks.

04First study · Proposal

90 participants, three arms, 12 weeks.

Randomize adults one to six months after stroke to standard rehabilitation, fixed BCI-FES, or adaptive AI-BCI-FES. Blinded change in FMA-UE is the primary outcome.

05Defensible data · Hypothesis

Record the full care loop—not merely model accuracy.

Link EEG confidence → stimulation decision → actual movement → fatigue and functional outcome over time. This becomes the basis for personalization and operations.

06Venture thesis · Assumption

Sell the care episode, not the device.

The first customers are rehabilitation and academic hospitals. Bundle assessment, personalized protocols, remote monitoring, and outcome reports per care episode.

Evidence map

The opportunity hidden in the evidence

+3.35 points

One-month between-group FMA-UE difference in a 17-center, 296-person trial

SMD 0.50

Pooled immediate effect across ten randomized BCI-FES trials

Not significant

No group difference in a small trial of severe paresis

Key variables

CST integrity, signal quality, and durability of effect

Evidence

Why now

Meta-analyses of randomized trials through 2025 suggest BCI rehabilitation can improve upper-limb function. Yet effect sizes and transfer to daily activity vary, and long-term multicenter evidence remains limited. The next question is not simply whether BCI works, but for whom, when, and how adaptation helps.

Multicenter randomized trial BCI-FES meta-analysis Clinical trial registry

Failure conditions and risks

Unstable signals

Pause automated stimulation when EEG quality drops and move to therapist-confirmation mode.

No clinical benefit

Judge success by functional recovery and daily activity—not classifier accuracy.

Over-personalization

Keep adaptation inside pre-approved bounds and log every change.

Weak business case

Test purchasing against therapist time, adherence, and outcome-reporting costs.

Bibliography

References

Established facts are separated from research and venture hypotheses that still require validation.

  1. 01

    Wang A, Tian X, Jiang D, et al. Rehabilitation with brain-computer interface and upper limb motor function in ischemic stroke: a randomized controlled trial. Med. 2024;5(6):559–569.e4.

    Seventeen centers and 296 participants; the between-group difference in one-month FMA-UE improvement was 3.35 points.PMID 38642555
  2. 02

    Brunner I, Lundquist CB, Pedersen AR, et al. Brain computer interface training with motor imagery and functional electrical stimulation for patients with severe upper limb paresis after stroke. J Neuroeng Rehabil. 2024;21:10.

    In 35 analyzed patients with severe paresis, the group difference was not significant; corticospinal tract integrity predicted recovery.PMID 38245782
  3. 03

    Ren C, Li X, Gao Q, et al. The effect of BCI-controlled FES training on upper-limb rehabilitation after stroke: a systematic review and meta-analysis. Front Hum Neurosci. 2024;18:1438095.

    Ten randomized trials and 290 participants showed a moderate immediate effect, while long-term evidence remained limited.PMID 39391265
  4. 04

    ClinicalTrials.gov. BCI-FES Therapy for Stroke Rehabilitation. NCT04279067.

    A registered trial comparing BCI-FES with dose- and intensity-matched standard physical therapy.NCT04279067
The next issue publishes every day at 9 AM KST.All issues