Bryan Chen
Notebook

NotebookEntry 01

The Biointerface Brief

Where brain-computer interfaces stood in September 2026, and why the tissue around the implant is still the hard part.

  • Research brief
  • 8 slides
  • 9 sources
  • Made with Claude
  • 26 September 2026

SEPT 2026 · RESEARCH BRIEF

The Biointerface
Brief

BCIs are scaling the hardware.
The tissue is still winning.

Bryan Chen · UCSD Microbiology

Slide 1BCI hardware is scaling fast, but the tissue around every implant is still the limiting problem. This brief covers where the clinical leaders stand, what the brain does to an implant, and the research bet I think is still open.

Where the clinical leaders are

Paradromics

Connexus · Connect-One study

Jun 17

First implant (2026), at U Michigan

Aug 26: FDA expanded the study to personal devices

Sep 14: a participant produced real-time speech

Neuralink

US · UAE · UK · Canada

26

People implanted as of July 2026, up from 21 in January

Blindsight holds FDA Breakthrough designation

“High-volume production” promised for 2026

Synchron

Stentrode · endovascular

6 of 6

COMMAND patients met the primary safety endpoint

16 electrodes delivered through a blood vessel, no open-brain surgery

Pivotal trial planned for 2026; $200M Series D (Nov 2025)

Sources: Paradromics news · trial trackers (Neuralink) · TechTimes, Jun 2026 (Synchron)

Slide 2Three companies lead the clinic, each with a different hardware bet. Paradromics went from first implant to real-time speech within three months, Neuralink is scaling its implant count, and Synchron avoids open-brain surgery by going through a blood vessel.

The brain sets the rules

People learned to steer an avatar in under an hour, but only when the decoder followed their brain’s existing activity patterns.

Mappings outside that neural manifold produced no learning at all.

Outside the manifold
No learning

Inside the manifold
Learned in under an hour

Busch et al., Nature Neuroscience, Jun 9 2026 (Yale)

Slide 3A Yale study is a reminder that the brain, not the decoder, sets the terms. People learned fast only when the decoder followed the brain's existing activity patterns; mappings outside that neural manifold were not learned at all.

What the tissue does to an implant

The foreign-body response (FBR) runs from minutes to months.

Minutes

Insertion injury

Capillaries torn, blood-brain barrier breached

Minutes

Protein coat

Plasma proteins adsorb onto the surface

Days

Microglia and macrophages

Swarm the surface; release cytokines and reactive oxygen species

Weeks

Astrocyte scar

A GFAP-dense sheath walls the implant off

Weeks to months

Signal loss

~40%

fewer neurons within 100 µm of stiff silicon; impedance rises

He et al., Int J Mol Sci 2026 (review) · Biran et al., Exp Neurol 2005

Slide 4Every implant sets off the same foreign-body response. Injury and a protein coat come within minutes, microglia and macrophages arrive within days, and an astrocyte scar walls the device off over weeks. Biran's 2005 study found about 40% fewer neurons within 100 µm of stiff silicon.

The publishing gap

4,992

BCI papers in 2025, 2.6× the 2017 count

2010

2017

2025

52

Papers on tissue response to electrodes in 2025, vs 51 in 2017

2010

2017

2025

Yearly counts, 2010 to 2025. Each chart is scaled to its own maximum.

Tissue papers per 1,000 BCI papers: 17.7 in 2013 to 2017, 12.8 in 2021 to 2025.

OpenAlex title and abstract search, pulled Sep 26 2026. Counts approximate.

Slide 5BCI publishing is up about 2.6-fold since 2017, while tissue-response papers stayed flat: 51 in 2017, 52 in 2025. Relative to the field, tissue work fell from 17.7 to 12.8 papers per thousand BCI papers. These are keyword counts from OpenAlex, so read them as a trend, not exact totals.

The open bet:
soft, porous, immunomodulatory

Soft and porous

pHEMA/GMA hydrogels with 40 µm pores, 4 weeks in rat brain

Less astrocyte encapsulation than solid rods

Softer gels, less pro-inflammatory macrophage polarization

New vessels and neuronal markers inside the pores

Dryg et al., Adv Healthc Mater 2026 (UW)

Active surfaces

A review mapping each strategy to the FBR step it targets

Antifouling coatings

Tethered neurotrophic factors

Drug-eluting interfaces

Immunomodulatory interfaces

He et al., Int J Mol Sci 2026

Measure it the same way

ML scoring of GFAP astrocytes around implants

The annotation strategy changes the result

Reproducibility is its own gap.

Melnikova et al., Int J Mol Sci 2026

Slide 6The open bet is an interface the tissue tolerates: soft, porous and immunomodulatory. Porous hydrogels reduced astrocyte encapsulation at four weeks in rats, active coatings can target specific FBR steps, and we still need a reproducible way to score the scar.

Three questions worth a lab

IN PRIORITY ORDER

1

Does soft and porous still win at 6 to 12 months, not 4 weeks?

2

Can a coating quiet microglia in week one: drug-eluting, or extracellular vesicles?

3

Can scar scoring become a reproducible standard?

Slide 7These are the three questions I would build a project around, in priority order. Durability comes first, because four weeks says little about a chronic implant; then early immune control; then a shared standard for scoring the scar.

Sources

Paradromics

Paradromics news, paradromics.com/news

Neuralink trial count

basenor.com (Jul 2026); Benzinga (Jan 2026)

Synchron

TechTimes (Jun 6 2026)

Neural manifold learning

Busch et al., Nature Neuroscience 2026, via Yale News (Jun 9 2026)

Porous hydrogels

Dryg et al., Adv Healthc Mater 2026, doi.org/10.1002/adhm.71522

FBR cascade and coatings

He et al., Int J Mol Sci 2026, doi.org/10.3390/ijms27156752

Scar scoring

Melnikova et al., Int J Mol Sci 2026, doi.org/10.3390/ijms27083524

Neuron loss near silicon

Biran et al., Exp Neurol 2005

Publication counts

OpenAlex API, openalex.org

Slide 8Every number in this brief traces to one of these sources. The publication counts come from OpenAlex title and abstract searches, so treat them as approximate.

About this brief

Claude searched news and papers from September 2026 and set the slides; every number traces to a source listed on the last slide. The speaker notes run under each slide, so the brief also reads fine on a phone.

The publication counts come from keyword searches of OpenAlex, so read them as a trend rather than exact totals. Entry 03 has the full series as an interactive chart, and Entry 02 animates the tissue response from slide 4.