Layer 3 — Voice Transaction (Healthcare)

Clinical Consent

Patient consent — spoken, witnessed, immutable.

Sesim Voice Transaction (Layer 3) extends from finance to healthcare consent: a patient authorizes a procedure, medication, or report access by speaking. The same NLP intent extraction, dual-confirmation, anti-spoof, voiceprint match, and immutable audit log apply — but with KVKK Art. 6 (special category data) and medical-ethics review built in. Phase 2 — partner-led pilot with hospitals, insurers, or public health platforms.

Phase 2 — partner-led pilot

Healthcare consent is in Phase 2 of our roadmap. We are looking for hospital + insurer or hospital + public-health partners for a pilot designed jointly with medical-ethics committees. Reach out if your organization wants to be early.

Discuss healthcare partnership →

Demo flow (illustrative)

A typical consent turn — illustrated only. No real audio is captured on this page; live execution requires a partner-led pilot.

1
Doctor reads consent
Procedure: appendectomy · Risk: anesthesia 0.3% · Date: tomorrow 09:00
2
Patient says
"I consent to the appendectomy"
3
ASR + intent
{ procedure: appendectomy, action: consent }
4
Dual-confirmation prompt
"Please say: 'Yes, code-cloud, I consent to appendectomy'"
5
Patient says
"Yes, code-cloud, I consent to appendectomy"
6
Cross-layer verdict
L1 stress 0.12 (low duress) · L2 voiceprint 0.89 · L3 anti-spoof 0.92 → CONSENT VALID
7
Immutable audit
Patient + doctor + procedure + voiceprint hash → blockchain audit · 7-year retention

Why voice consent is hard

Patient may be in pain, on medication, or anxious — coercion / cognitive-load detection is essential.
Family members may be present — distinguishing patient voice from family voice requires voice biometric.
Multilingual patients (Turkish + Kurdish + Arabic + English) — code-switching ASR.
Consent must be informed — the system must verify the patient repeated the procedure name correctly.
Audit must be immutable — disputes (medical, legal) require evidence chain.
Patient may revoke consent later — right-to-erasure plus medical record retention rules collide.

Sesim approach

Voice biometric (L2): patient enrolled at admission; voiceprint match on consent.
Coercion detection (L1): stress + duress patterns flag suspicious consent for medical-ethics review.
Dual-confirmation: patient repeats procedure name + dynamic random word — no surprise consent.
Anti-spoof: TTS clones and recordings rejected.
Audit log: voice intent + voiceprint hash + L1+L2 scores + timestamp + medical-record link.
KVKK m.6 lawful basis: explicit patient consent + medical-ethics committee approval + DPA addendum (separate from financial L3).

Pilot design (Phase 2)

Partner: hospital + insurer (or hospital + public health) consortium.
Scope: low-risk elective procedures or consent for medication / lab access — not emergency.
Duration: 16–24 weeks (medical-ethics + KVKK + tech + UX rounds).
KPI: consent capture success ≥ 95%, voice-revocation latency ≤ 5 min, dispute resolution ≤ 14 days.
Audit: hospital + insurer + Sesim three-way reconciliation.

Pilot at a glance (Phase 2 roadmap)

Status

Phase 2 — partner-led, not active

Duration

16–24 weeks once partners signed

Fee

$50–80k creditable (TBD with first partner)

Compliance

KVKK m.6 (special category), medical-ethics committee approval

Phase 2

$300–600k yearly per hospital network (TBD)