Novel Biotech · limen Step 5 research-target definition · producer candidate, not accepted · 2026-08-21

Which limen research target is strongest — and why none is a product

We compared seven problem classes against the inspected human and animal evidence and against lower-risk alternatives. One class earns the next research question. Nothing here selects a product, an implant, or a program.

Words used in this report

Direct natural ENG

Electrical activity recorded from peripheral-nerve axons themselves. Not muscle activity, not an evoked response.

Muscle EMG / surface EMG

Electrical activity from muscles — including RPNI grafts. A different signal from nerve recording, never interchangeable with it.

Forward route

CNS → peripheral nerve → recording interface → computer.

Return route

Computer → stimulation interface → afferent peripheral nerve → CNS.

Simultaneous duplex

Usable natural nerve recording during stimulation at the same interface, with timing shown explicitly. Unknown in humans today.

Lower-risk comparator

An external, removable, or temporary alternative (for example optimized surface-EMG control). A category label, not a safety finding.

R0

A planned offline, synthetic-only, non-actuating software simulator. It does not exist yet.

What was researched

We re-inspected the accepted evidence register: human nerve recording and stimulation studies, muscle-based interfaces, long-duration implant records, failures and adverse events, lower-risk comparators, official programs, patent publications, and regulatory records — 84 registered sources in total. We then asked, for each of seven problem classes, whether an invasive nerve interface could add a function that an optimized external alternative cannot deliver on a matched population, endpoint, and duration. [source register]Private repository source

What was found

• The exact combination was not established. Human studies have shown nerve recording, nerve stimulation, and muscle-based bidirectional control — some for years. No inspected source shows a chronic, fully implanted system combining stable direct natural ENG with afferent return. [S5-CR-002]Private repository source
• The longest inspected direct-recording record contains an early failure warning. Six of seven arrays lost functional recording electrodes within about two months; one participant kept recording past day 500. The mechanism remains unknown. [B01]
• No matched comparison exists in the bounded corpus. No eligible study compared an invasive route against an optimized lower-risk alternative on the same population, function, endpoint, and duration. That gap stays open; it is not proof either way. [comparator audit]Private repository source
• The same-participant caveat holds everywhere. Studies that recorded and stimulated in the same person did not demonstrate simultaneous same-interface operation, and the one closed-loop task used mixed nerve-plus-muscle signals. [B00]

What remains unknown

• Why recording contacts fail early — device, access route, biology, or care — is unknown. [S5-CR-004]Private repository source
• Whether usable natural nerve recording during stimulation at the same interface is achievable in humans: unknown. [S5-CR-010]Private repository source
• Whether any invasive route beats an optimized external one on a matched comparison: unknown in all seven classes. [S5-CR-060]Private repository source
• Long-term selectivity, calibration life, reversibility, repairability, upgradeability: open or unknown.
• Whether a living-tissue intermediary adds anything separable: hypothesis only.

Which research target is strongest

PRIMARY  S5-T1  Forward-route signal stability
         population : adults with unilateral transradial limb loss
         question   : does direct nerve recording keep its prespecified
                      function across a declared multi-month window?
         comparator : optimized surface-EMG control (removable)
         kill       : comparator matches the function; or candidate
                      misses its endpoint; or contact loss proves
                      structural in the declared window

BACKUP   S5-T2  Return-path encoding stability
         population : adults with upper-limb loss using myoelectric prostheses
         question   : does stimulated sensation stay stable vs external feedback?
         comparator : optimized vibrotactile / electrocutaneous feedback

Both targets are research-evidence targets: they exist to close open questions. Neither clears an evidence threshold today, and neither may be called promising, feasible, or selected as a product direction.

Why it is not a product

• The decisive questions above are unanswered, and the observed failure mode is severe.
• No eligible head-to-head comparison exists anywhere in the inspected corpus.
• The R0 simulator does not exist; no reference vectors exist; ten backend technology decisions are unselected.
• By standing boundary: no development program exists, nothing is available, and no human testing is contemplated.

What must happen next

1. Independent QA audits this package; the Integrator accepts or returns it.
2. Science folds the demanded scenario classes into versioned contracts; Backend resolves its ten technology decisions.
3. Synthetic reference vectors are created and accepted — still no biological data.
4. Only then: implement and validate the offline R0 simulator. A new non-biological bench comes later, if ever, on separate authorization.

What we know / What we do not know / What must happen next