AI Scribe for Psychiatrists Mental Status Exam

AI Scribe for Psychiatrists: The Mental Status Exam Documentation Standard

Merry AI · Thoughtfully curated clinical briefs.

• Jump to: The Loaded Labor Model
• Read next: Clinical Logic & Audit Defense
• Coding trace: ICD-10 Documentation Standards
• The wedge: The Group-Block Blind Spot
• Delivery layer: Chrome Extension DOM Injection
• Orchestration: Clinical Intelligence Layer

TL;DR for busy clinicians: A psychiatric AI scribe should defend the MSE, not just draft it.
The measurable value sits in $15,600+ recovered CPT G2211 revenue and 2.1+ hours saved daily.
The unsolved problem is PHP/IOP group blocks — one session, ten individualized billable notes.
Merry AI splits group encounters and injects finalized notes into closed EHRs via Chrome Extension.

The mental status exam remains the psychiatric equivalent of a physical exam finding — the clinician's directly observed evidence, not a patient-reported summary. Merry AI treats the MSE as a defensible record first and a draft second.

Documentation failure at the MSE converts genuine clinical work into unsupported claims and audit exposure. In multi-site PHP/IOP settings, that exposure compounds across every attendee in every group block, which is where Merry AI concentrates its logic.

Section 1: The Loaded Labor & Denominator Model

CLINICAL UPDATE 2026: Revised for new CMS CPT G2211 standards, SB 1120 compliance, and FHIR interoperability.

Cost should be read as a labor-denominator ratio, not a monthly subscription line. The correct comparison is the staff member the scribe offsets.

A medical assistant carries roughly a $48,000 fully loaded annual cost against a $35,000 base wage, once benefits and overhead are added.

Merry AI Pro runs $648 per year ($54/month annual), which lands at approximately 1.3% of that loaded labor line.

Recovered revenue reframes the arithmetic entirely: $15,600+ annually through CPT G2211 complexity capture, plus 2.1+ hours returned daily per provider within published JAMA/NEJM benchmark ranges.

Review the plan structure directly at Merry AI Practice Partner Plans before comparing against your current staffing model.

Section 2: Clinical Logic & Audit Defense

Consider a multi-site PHP/IOP psychiatrist reviewing a 3-hour group and medication-management block with 10 attendees. A single transcript for that block is a clawback waiting to be filed.

Merry AI splits that encounter into 10 individualized notes, one per patient, each anchored to that patient's stated symptoms.

• Each note preserves observed symptoms, MSE findings, and PHQ-9 / C-SSRS risk language specific to that attendee.
• Psychotherapy add-on support attaches for CPT 90833 or 90836 where time is separately documented.
• Finalized notes inject into Kipu through the Chrome Extension DOM workflow, with no manual copy-paste step.

The clinical outcome is individualized documentation that avoids the cloned-note pattern Joint Commission reviewers flag on sight. Documentation integrity here is the anchor: every note must read as a distinct clinical event.

Human-attested clinical metrics anchor the defense — LVEF %, ROM degrees, DSM-5-TR criteria, and numeric symptom scores are framed as clinician-verified findings, not model inferences. See how the Psychiatry Mental Status Exam Workflow preserves this attestation chain.

This directly reduces clawback exposure under California SB 1120 utilization-review rules and NCCI Modifier 25 audit logic.

Section 3: Clinical Taxonomy — ICD-10 Documentation Standards

Accurate coding follows accurate MSE language, never the reverse. The finding is documented first; the code is a downstream consequence.

• F33.1 documents recurrent, moderate major depressive disorder, supported by severity findings in the note body.
• R45.851 captures suicidal ideation and must trace back to explicit C-SSRS language, attested by the clinician.
• Each code requires a human-attested clinical finding rather than a generated inference standing alone.

Reference the official taxonomy directly: F33.1 — Major depressive disorder (ICD-10-CM) · recurrent (ICD-10-CM) · moderate; R45.851 — Suicidal ideations (ICD-10-CM).

Section 4: The Group-Block Blind Spot Most Scribes Never Solved

Most psychiatric scribes generate one transcript per session and one note per session. That assumption holds in a solo outpatient office and collapses in PHP/IOP.

The PHP/IOP reality differs entirely: a single 3-hour block can hold up to 10 individually billable encounters under one clinician's supervision.

• The unaddressed risk is not transcription accuracy — it is cross-attendee note-cloning, where the same MSE language repeats across records.
• Cloned notes trigger Joint Commission warnings and NCCI Modifier 25 clawbacks during retrospective review.
• Competitor coverage stops at MSE, SI/HI, and polypharmacy for single sessions and never addresses the group block.
• The workflow wedge is per-attendee splitting with individualized risk language preserved in each output.

The Path Recovery TN case study demonstrated this directly: one group block produced ten discrete, defensible notes without a manual re-write step.

The underlying evidence base supports structured, individualized capture — see NCBI.NLM.NIH Clinical Research and the broader NIH National Library of Medicine Research archive.

Section 5: Chrome Extension DOM Overlay & EHR Field Injection

Browser-native architecture requires zero IT provisioning, server standup, or API credential exchange. The extension runs in the same tab as the chart.

• A DOM overlay reads and writes directly to the visible EHR fields the clinician already sees.
• Closed EHR compatibility extends to systems without open APIs, including Kipu.
• Field injection places finalized notes into the correct chart section automatically.
• PHP/IOP group note-splitting outputs each individual note to its own patient record.

Comparable field-injection behavior is documented in the SimplePractice AI Scribe Integration reference.

Section 6: Clinical Intelligence Layer — Closed-Pilot Orchestration

The Practice Partner plan runs $149/month and opens the closed-pilot Clinical Intelligence Layer, where five outpatient practices are selected weekly for direct solutions engineering.

• Pre-visit orchestration surfaces prior MSE, medication history, and open risk items before the patient is seen.
• During-visit capture structures the encounter into MSE, assessment, and plan sections in real time.
• Post-visit finalization applies coding trace and injects each note into the EHR record.

The same layer carries SB 1120 and NCCI audit shields, mapped to the plans listed at Merry AI Practice Partner Plans.

The measure of a psychiatric scribe is audit survival, not draft speed.