MedLog AI Clinical Scribe

Freed, Heidi, Suki, and Doximity Scribe all upload patient audio to their cloud. Abridge is under active class-action litigation in California for unconsented recording in Sharp HealthCare clinics. The cloud-scribe category has a structural privacy problem. MedLog AI sidesteps it entirely with Apple Foundation Models running on-device.

THE 60-SECOND ENCOUNTER
1. Tap the big mic button. Speak the encounter. Patient and clinician audio captured locally via on-device SFSpeechRecognizer.
2. Tap Stop. Foundation Models structures the audio into SOAP sections: Subjective, Objective, Assessment, Plan.
3. ICD-10-CM code suggestions appear next to the Assessment section, 130+ code database, keyword + description matching with confidence scores.
4. Review, edit, sign.
5. Export: PDF (for hand-off), FHIR R4 (for Epic / Athenahealth / Elation upload), or plain text (for clipboard paste).

FHIR R4 EXPORT, NATIVE
Patient resource (encrypted PHI fields), Encounter resource (with state-machine status), Condition resource (ICD-10-CM codes), Observation resource (vitals + LOINC codes), DocumentReference resource (the signed note).

No other solo-clinician scribe exports clean FHIR R4. Freed exports plain text. Heidi exports plain text + Word. Only MedLog speaks the standard your EHR's import path actually wants.

WHO IT IS FOR
Solo NPs, PAs, MDs, DOs running independent practices or urgent-care moonlight shifts. Dentists charting between patients.

PRIVATE BY DESIGN
Foundation Models on-device. No cloud LLM call. Audio never recorded to disk. Transcription happens in memory. Patient records encrypted at rest via CryptoKit and Secure Enclave. Local Face ID lock. App-level passcode optional. Append-only audit log (exportable for your records). No analytics. No advertiser SDKs. No telemetry that includes PHI.

APPLE WATCH SUPPORT (v1.1)
The bundled MedLogAIWatch target is in development for v1.1. Quick voice capture from the wrist mid-rounds. Watch UI is not shipped in this version.

HONEST DISCLOSURES
MedLog AI is a documentation tool. It does not provide medical advice or replace clinician judgment. ICD-10 code suggestions require clinician review before submission to payers. Apple Foundation Models inference quality varies with audio clarity and encounter complexity. This app is not a substitute for compliance review by your practice's privacy officer.

Support: [email protected]

Terms of Use (EULA): https://www.apple.com/legal/internet-services/itunes/dev/stdeula/
Privacy Policy: https://nexasphere.io/products/iphone/medlog-ai/privacy