Example Governed Decision Record, Neurotech

GOVERNED DECISION RECORD

Record ID:           omega-record-2026-05-12-neuro-001
Decision date:       2026-05-12 02:17 UTC
Recording authority: Dr S. Iwasaki (on-call neurotechnology
                     clinician, implanted BCI service)
Domain:              Implanted brain-computer interface
                     (closed-loop adaptive stimulation)
Schema Version:      omega/1.0
Composition:         substrate_native_records →
                     omega_records (canonical envelope)

CONTEXT

Implanted closed-loop deep brain stimulation device
(model NX-Adaptive-2, firmware 4.7.1) operating under
the institutional Adaptive Neurostimulation Service
care pathway (ANS-CP-2024-v2.1). Patient is a 58-year-old
adult with medication-refractory essential tremor,
implanted 14 months prior, stable on closed-loop
adaptive policy since 2026-02-04.

Declared adaptive stimulation envelope:
- Target biomarker: ventral intermediate nucleus
  beta-band power, 13-30 Hz
- Stimulation parameter ranges: amplitude 0.5-3.5 mA,
  frequency 130-180 Hz, pulse width 60-120 µs
- Declared biomarker envelope: tremor-correlated
  beta-band power within ±2 standard deviations of
  the patient's seven-day rolling baseline
- Declared termination conditions: sustained biomarker
  drift beyond envelope for >15 minutes, any stimulation
  parameter excursion beyond ranges, any patient-reported
  adverse event via paired handset, any impedance
  excursion suggesting lead fault
- Pre-execution expectation: nocturnal biomarker
  trajectory between 01:00 and 04:00 UTC should remain
  within the rolling baseline envelope

Operational authority structure:
- Implanting Consultant Neurosurgeon: Mr H. Cavendish
  (off-site, on-call escalation)
- Movement Disorders Consultant: Dr P. Adeyemi (off-site
  overnight; responsible clinician for stimulation policy)
- On-call neurotechnology clinician (recording
  authority): Dr S. Iwasaki (remote device console)
- On-device adaptive policy: ML inference model deployed
  as parameter-adjusting layer within declared envelope
  per institutional ANS-CP-2024-v2.1
- Manufacturer field engineer: notified for record only
- Patient: notified at 02:21 UTC via paired handset
  confirming reversion to safe-default programme

Frameworks cited:
- EU Medical Device Regulation 2017/745 (MDR), implantable
  active medical device requirements, post-market
  surveillance, vigilance reporting
- IEC 60601-2-10:2012, particular requirements for
  basic safety and essential performance of nerve and
  muscle stimulators
- ISO 14971:2019, application of risk management to
  medical devices
- ISO/IEC 42001:2023, AI management systems
- EU AI Act Article 12, record-keeping requirements
  for high-risk AI systems
- Institutional ANS-CP-2024-v2.1, Adaptive
  Neurostimulation Service Care Pathway (closed-loop
  envelope authority, clinician revert authority,
  safe-default programme protocol)

DECISION

Revert implanted device to clinician-validated
safe-default open-loop stimulation programme
(SAFE-DEF-04, fixed amplitude 1.8 mA, frequency
145 Hz, pulse width 90 µs) at 02:17 UTC. Suspend
adaptive policy updates and freeze on-device model
state. Maintain telemetry. Route question to morning
device safety review board at 09:00 UTC for resumption,
re-tuning, or programme termination decision.

REASONING

1. On-device adaptive policy logged sustained biomarker
   drift at 01:58 UTC, with beta-band power tracking
   +3.4 standard deviations above the seven-day rolling
   baseline for 19 continuous minutes (envelope is
   ±2 SD over >15 minutes).

2. Device alert raised to remote monitoring console at
   02:13 UTC after on-device confirmation window.
   Adaptive policy confidence band: 0.78 on drift
   detection, 0.39 on cause classification (model
   declined to commit between lead-tissue interface
   change, baseline non-stationarity, and a possible
   intercurrent sleep-architecture disturbance).

3. On-call neurotechnology clinician reviewed alert at
   02:15 UTC via remote device console. Telemetry
   review confirmed sustained drift, no impedance
   excursion, no stimulation parameter excursion within
   the envelope, and no adverse event flag from the
   paired patient handset.

4. Alternative options considered:
   - Allow adaptive policy to continue exploring within
     declared envelope: rejected. Declared termination
     conditions require revert on sustained biomarker
     drift beyond envelope; this is non-discretionary.
   - Hold current adaptive parameters and observe:
     rejected. Holding leaves the patient on a parameter
     setting the adaptive policy reached under
     out-of-envelope biomarker conditions; clinical
     interpretability of that setting is degraded.
   - Disable stimulation entirely: rejected. Sudden
     cessation in chronic DBS for essential tremor
     risks rebound symptoms and is not warranted in the
     absence of an acute safety event.
   - Escalate to Movement Disorders Consultant
     immediately: deferred. Revert-to-safe-default is
     within on-call clinician authority under
     ANS-CP-2024-v2.1; consultant review is scheduled
     for morning board.

5. Decision: revert to SAFE-DEF-04, suspend adaptive
   updates, freeze on-device model state, escalate to
   morning review board. Decision aligns with declared
   termination conditions and ANS-CP-2024-v2.1
   adaptive-layer authority limits (on-device policy
   may adjust parameters within envelope; humans decide
   reverts; on-call clinician authority permits revert
   to a pre-validated safe-default programme without
   consultant escalation; consultant consulted in
   morning review).

EVIDENCE AND ASSUMPTIONS

Direct observations:
- On-device drift detection record at 01:58 UTC
  (biomarker trace retained, model version retained,
  on-device policy hash retained)
- Clinician telemetry review at 02:15 UTC
- Stimulation parameter log: within declared envelope
  throughout the drift window (no amplitude, frequency,
  or pulse-width excursion)
- Impedance log: stable across both contacts; no fault
  signature
- Patient handset event log: no adverse-event flag,
  no patient-initiated programme change
- Seven-day rolling biomarker baseline retained

Assumptions made:
- On-device adaptive policy training distribution
  includes the class of biomarker drift observed
  (chronic DBS for essential tremor under nocturnal
  conditions)
- Telemetry stream accurately represents on-device
  state (no radio link artefact, no sampling drift)
- Reversion to SAFE-DEF-04 preserves a clinically
  acceptable symptomatic state for the patient
  overnight without introducing new risk
- Freezing on-device model state preserves the policy
  parameters at time of revert for morning forensic
  review

What is unknown at time of recording:
- Whether the biomarker drift indicates a lead-tissue
  interface change, a baseline non-stationarity (e.g.,
  intercurrent sleep-architecture disturbance), or a
  policy distribution-shift false alarm
- Whether the adaptive policy parameters reached prior
  to revert remain clinically appropriate for daytime
  use once cause is identified
- Whether the underlying cause, once identified, will
  prompt re-validation of the adaptive policy class
  across the cohort served by ANS-CP-2024-v2.1
- Whether the patient will experience symptomatic
  difference on SAFE-DEF-04 sufficient to warrant
  expedited rather than scheduled morning review

AUTHORITY

On-device adaptive policy authority: parameter
adjustment within declared envelope only. The model
has no autonomous revert authority and no authority
to exceed declared parameter ranges under
ANS-CP-2024-v2.1. Drift event was routed to on-call
clinician per declared institutional protocol.

On-call neurotechnology clinician authority: revert
to a pre-validated safe-default programme, suspend
adaptive updates, and freeze on-device model state
without consultant consultation. Escalate to
Movement Disorders Consultant at morning review.
Recording authority for this record.

Movement Disorders Consultant authority: scheduled
to convene morning device safety review board at
09:00 UTC. Authority to direct resumption,
re-tuning, or programme termination. Authority to
require re-validation of the adaptive policy class
if cause analysis warrants.

Implanting Consultant Neurosurgeon: notified for
record; authority required only if lead-fault or
explant question arises.

Manufacturer field engineer: notified for record;
no operational authority; vigilance reporting
obligation under MDR Article 87 assessed and
determined not yet triggered at this stage.

CONSTRAINTS CHECKED

- On-device adaptive policy operating within declared
  envelope authority (✓)
- Revert-to-safe-default within on-call clinician
  operational scope per ANS-CP-2024-v2.1 (✓)
- SAFE-DEF-04 is a clinician-validated programme
  previously demonstrated tolerable for this patient
  during the post-implant titration phase (✓)
- Stimulation parameters at revert within IEC 60601-2-10
  particular safety requirements (✓)
- Patient notified via paired handset and acknowledgement
  recorded (✓)
- Telemetry retained at full resolution for forensic
  review (✓)
- On-device drift event, biomarker trace, model
  version, policy hash, and confidence band recorded
  for ISO 42001 audit trail (✓)
- EU AI Act Article 12 record-keeping requirements
  satisfied for high-risk AI system intervention (✓)
- MDR Article 87 vigilance reporting threshold
  assessed and determined not triggered at this
  stage; re-assessment at morning review (✓)

DISPUTES

None at time of recording.

Pending: morning device safety review board may
dispute the revert decision if cause analysis
identifies the drift as a benign baseline
non-stationarity (e.g., sleep-architecture
artefact). Such a dispute would be recorded as a
separate governance event with the review board's
reasoning, the original revert reasoning retained
unchanged, and the policy implications (whether the
declared biomarker envelope was appropriately
specified for nocturnal conditions) flagged for
care-pathway review.

CONSENT

Patient consent to closed-loop adaptive stimulation
was recorded at the post-implant titration consent
event (2026-02-04). That consent expressly included
the safe-default revert pathway and the on-call
clinician's authority to invoke it without contacting
the patient ahead of the action. Patient was notified
of the revert at 02:21 UTC via paired handset and
acknowledged the notification at 02:23 UTC. No
real-time consent decision was clinically required
for this action under the consented care pathway.

HARM TRACE

Counterfactual if on-device drift alert had been
ignored:
- Continued adaptive parameter exploration under
  out-of-envelope biomarker conditions
- Possible exposure of the patient to stimulation
  settings the adaptive policy would not have reached
  under declared conditions
- Detection delayed to the next scheduled telemetry
  review, with adaptive policy state irretrievable at
  the granularity preserved by an immediate freeze

Counterfactual if on-device adaptive policy had not
been deployed:
- Patient remains on prior open-loop programme with
  known symptomatic trade-offs (greater off-time,
  higher stimulation duty cycle)
- No automated nocturnal drift detection; biomarker
  trajectory reviewed only at scheduled clinic visits

Counterfactual if revert had been delayed for further
investigation:
- Unknown; depends on actual underlying cause
- If lead-tissue interface change: harm scales with
  continued operation on parameters tuned against a
  changed substrate
- If baseline non-stationarity: harm bounded by the
  duration of the non-stationary period
- If policy distribution-shift false alarm: no harm,
  but the patient would have spent the night on
  adaptive parameters reached under out-of-envelope
  conditions, raising governance questions about
  envelope specification for nocturnal use

Counterfactual if on-device policy had been granted
autonomous revert authority:
- Revert would have occurred at 01:58 UTC rather than
  02:17 UTC (nineteen minutes earlier)
- Revert would have been recorded as autonomous
  on-device action rather than human clinician
  decision, outside current institutional authority
  structure and outside the consent recorded at the
  post-implant titration event
- Substantive trade-off: speed of intervention versus
  authority appropriate to model confidence band
  (0.39 on cause classification is below threshold
  for autonomous device action under ANS-CP-2024-v2.1)

OUTCOME

Device reverted to SAFE-DEF-04 at 02:17 UTC.
Adaptive policy updates suspended; on-device model
state frozen for forensic review.
Telemetry maintained at full resolution.
Patient notified at 02:21 UTC; acknowledged 02:23 UTC.
Morning device safety review board convened at
09:00 UTC.
Cause analysis pending.

Resumption / re-tuning / programme termination
decision pending morning review board outcome
(separate governance record).

PROVENANCE

Content hash:          [sha-256: pending review board
                       outcome, then sealed]
Previous record hash:  [adaptive policy commission
                       record 2026-02-04-neuro-comm-027]
Schema validation:     passed (omega-contracts v0.2.2)
Cryptographic seal:    valid (interim seal pending
                       outcome record)

Linked records:
- Adaptive policy commission record
  (2026-02-04-neuro-comm-027)
- On-device policy training attestation
  (2026-01-15-nx-att-009)
- Institutional ANS-CP-2024-v2.1 record
- Patient consent record, post-implant titration
  (2026-02-04-neuro-cons-027)
- Implant record and lead impedance baseline
  (2025-03-11-neuro-impl-019)
- Forthcoming: morning device safety review board
  outcome record

This example record is a fictional implanted
brain-computer interface scenario, composed across
the OMEGA substrate record schema for illustration.
No real patient, clinician, institution, or device
platform is depicted.