AI Scribe for Bariatric & General Surgery

Specialty Documentation Architecture for General Surgery

Clinical documentation in bariatric surgery is not a single note but a longitudinal evidentiary chain assembled across referral intake, multidisciplinary preoperative evaluation, a supervised weight-management period, and a consolidated medical-necessity packet. Merry AI is positioned as a documentation-retrieval and drafting function: it references a patient's historical encounter data to help auto-populate the preoperative trial packet, but it never replaces clinician validation, payer-specific criteria review, or the surgeon's attestation that the record supports medical necessity.

Because payer rules are not uniform, the anchor requirement of "six months" must be treated as a payer-configurable rule rather than a universal federal standard. Illinois Medicaid requires six consecutive months of medically supervised weight loss within the year preceding the prior-approval request, while CMS billing guidance describes a physician-supervised program of at least four consecutive months within the prior twelve months, with monthly weight, BMI, diet, and physical-activity documentation. Commercial policies such as Priority Health require at least six office visits with distinct diet and exercise components. The rules engine encodes these variants so a draft never overstates compliance.

The safest implementation model is source-linked auto-population. Each generated field retains its source encounter date, source document and author, original wording or measured value, extraction confidence, and whether the value was directly documented, calculated, inferred, or manually entered. This provenance discipline is what makes a six-month packet defensible on complex Recovery Audit Contractor review, which CMS has approved for bariatric procedures 43770, 43644, 43645, 43845, 43846, 43847, and 43775.

Referral Intake and Eligibility Capture

Eligibility documentation begins with the anthropometric foundation: height, weight, BMI, weight trajectory, and obesity duration. Medicare considers surgical management reasonable and necessary for beneficiaries with a BMI at or above 35, at least one obesity-related comorbidity, and prior unsuccessful medical treatment of obesity. Merry AI extracts each of these fields from vitals feeds rather than narrative mentions, computing BMI only from verified height and weight in reconciled units, and preserving the original values so a pounds-versus-kilograms mismatch never silently corrupts the record.

Comorbidity capture during intake consolidates diabetes, hypertension, dyslipidemia, obstructive sleep apnea, GERD, and degenerative joint disease alongside prior bariatric or abdominal surgery, anti-obesity medication history, and a structured account of prior weight-loss attempts with outcomes, duration, and reasons for discontinuation. Peer-reviewed medical-necessity frameworks indexed at the NCBI PubMed Central archive reinforce that contemporaneous comorbidity documentation is the strongest predictor of first-pass authorization approval.

The Clinical Logic Matrix

The following matrix maps each bariatric diagnostic framework to the required clinical data points and the corresponding billing or authorization evidence. It is the operational core of how Merry AI routes extracted findings into payer-approved packet sections.

Supervised Weight-Management Period Logic

The supervised period is where false-completion risk concentrates. Merry AI distinguishes four encounter classes: a visit where weight was merely measured, a visit where diet or exercise counseling was documented, a visit that actively addressed obesity management, and a visit that satisfies the specific payer's qualifying criteria. A generated statement such as "patient completed six consecutive months of supervised weight loss" is prohibited unless the rule engine confirms the required dates, visit count, and documentation elements, and a clinician verifies the result.

Gap detection runs continuously across the configured look-back window, flagging missing monthly weights, incomplete counseling documentation, absent dietitian visits, non-consecutive months, or an insufficient number of qualifying encounters. Each gap is surfaced to the bariatric coordinator with a direct link back to the timeline so missing evidence can be requested before submission rather than discovered after a denial.

Medical-Necessity Packet Generation

The consolidated packet assembles a complete history and physical, current height, weight, BMI, and weight trajectory, obesity-related comorbidities, prior weight-loss attempts, medical and surgical history, medication history, visit-by-visit supervised weight-management records, dietary and behavioral evaluations, relevant laboratory, imaging, sleep-study, and endoscopy results, operative-risk assessment, the proposed procedure with rationale, preoperative and postoperative treatment plans, procedure and diagnosis codes where requested, and the signed letter of medical necessity. MassHealth guidance, for reference, explicitly enumerates these same elements.

Every field in the generated packet is labeled either "AI-extracted — review required" or "clinician verified," so the surgeon can never mistake an unreviewed extraction for confirmed fact. The interface presents a chronological timeline, opens each source note from the packet field, and lets the clinician accept, edit, reject, or exclude any extracted value, mark an encounter as qualifying or nonqualifying, and resolve contradictions such as conflicting weights or late-signed notes before attestation. You can Access Specialty Prompts at templates.scribing.io to load these packet scaffolds directly.

Compliance, Validation, and Governance

Because the workflow handles protected health information end to end, deployment requires a signed HIPAA Business Associate Agreement, role-based least-privilege access, multifactor authentication, encryption in transit and at rest, immutable audit logs for viewing, exporting, editing, and transmitting records, and defined retention and breach-response procedures. CMS bariatric structural standards specifically reference HIPAA-compliant documentation of multidisciplinary committee activity, so provenance and integrity controls are not optional add-ons.

Before clinical deployment, the system is validated against a representative bariatric chart set spanning complete six-month records, incomplete records, gapped records, duplicate encounters, conflicting weights, late-signed notes, scanned outside records, and multiple payers. The governing safety metric is the false-positive completion rate — the proportion of packets incorrectly represented as satisfying a payer's duration or documentation requirements — alongside field-level extraction precision and recall, correct qualifying-encounter identification, unsupported-statement rate, clinician correction frequency, and authorization denial or resubmission rate. To pressure-test these controls against your own chart mix, Schedule a 15-Minute Specialty Workflow Audit. The central principle is unchanged: Merry AI retrieves and organizes historical evidence, but the record remains clinically and legally attributable to the treating organization.