[Eeglablist] PSI discussion draft and synthetic EEG results (research only)

[נתי שטרן]

Dear colleagues,

I am sharing the updated PSI discussion draft, "Reframing the
Psychiatric State Index Through Clinical Assessment and Measurement
Validation" (14 October 2026):
https://urldefense.com/v3/__https://github.com/netanelcyber/psi-qeeg-bis/blob/18e22b6dd4ddc8747b6c2034cc7f457ec1aff28f/docs/research/PSI_CLINICAL_PERSPECTIVE.md__;!!Mih3wA!ARYPdYz3Ra3_a3PPOPiAGqTdZscErQ1gC4LmZoO29pqz4q5w252Xc8PDT2ScbSZ3Ybtk0M6Er8PGx9irsDo$ 

This is an unpublished, non-peer-reviewed perspective and proposed
validation protocol. Psychiatric review of the protocol remains
pending. It places clinical assessment at the reference standard and
excludes scalp gamma and scalp theta–gamma phase-amplitude coupling
(PAC) from the proposed primary human analyses.

Separate software demonstration — synthetic data only:
The BTBR-labelled example uses four simulated animal identifiers,
three simulated days and two frontal-labelled channels sampled at 256
Hz. These are not real animal recordings. The generator deliberately
inserts a theta/gamma change during the 180 seconds preceding
simulated grooming onsets; the 180-second setting is not a
demonstrated warning time.

The preliminary author-reported feature comparison for that simulation
gives the following descriptive Cohen's d values:
- Gamma coherence, L–R: 4.44 (reported means approximately 0.78
baseline versus 0.96 pre-event).
- Relative gamma, R: 1.48.
- Beta power, L: 1.41.
- Relative gamma, L: 1.35.
- Gamma power, R: 1.22.
- Gamma power, L: 1.19.
- Relative theta, L: 0.52.

These feature contrasts have not been independently reproduced for
this update; no cluster-aware uncertainty estimates accompany the
supplied summary. They concern correlated synthetic windows, not
independent biological observations. The archived simulation output
also contains window-based model metrics, which must not be treated as
validated event-level warning performance.

Interchannel gamma coherence is distinct from theta–gamma PAC. It does
not establish cross-frequency coupling, bilateral cortical "locking",
or a brain region carrying clinical risk. The deliberately simulated
signature cannot establish a biological mechanism or real-world
prediction.

Neither the draft nor this demonstration establishes diagnostic
discrimination or prospective prediction of autistic meltdown or
psychosis. The separate demonstration branch does not implement or
validate the proposed human protocol. The simulation results have not
been peer reviewed.

Archived simulation output:
https://urldefense.com/v3/__https://github.com/netanelcyber/psi-qeeg-bis/blob/ba5f079eb6295e725dd52caf84924468e6399c21/outputs/btbr_prediction/prediction_results.json__;!!Mih3wA!ARYPdYz3Ra3_a3PPOPiAGqTdZscErQ1gC4LmZoO29pqz4q5w252Xc8PDT2ScbSZ3Ybtk0M6Er8PG_NTSPpw$ 

Generator and provenance:
https://urldefense.com/v3/__https://github.com/netanelcyber/psi-qeeg-bis/blob/ba5f079eb6295e725dd52caf84924468e6399c21/examples/btbr_demo/run_prediction.py__;!!Mih3wA!ARYPdYz3Ra3_a3PPOPiAGqTdZscErQ1gC4LmZoO29pqz4q5w252Xc8PDT2ScbSZ3Ybtk0M6Er8PGC9bWpKw$ 

Methodological comments on clinical labels, independent
EMG/EOG/movement controls and participant-level validation would be
welcome. No participant recruitment is open.

Best regards,
Netanel Stern