Endocrinology AI scribe 2026: diabetes visits, GLP-1 documentation, thyroid, and the insurance-driven note reality

May 8, 2026 · 7 min read

Endocrinology in 2026 is documentation-heavy in a different way from cardiology or surgery. The work isn't procedural; it's longitudinal. Each visit involves trend data (A1C / CGM / fructosamine / TSH series), insurance-driven prior auth gates (GLP-1, ozempic, mounjaro for non-diabetic obesity), and behaviorally-loaded counseling (insulin titration, dietary, exercise, weight). General AI scribes capture the conversation but rarely capture what insurance reviewers actually look at when reviewing a chart.

The 2026 endocrinology-aware AI scribe stack handles four things general scribes miss: trend-data interpretation, prior-auth-defensive note structure for GLP-1 / weight-loss meds, ICD-10 specificity for diabetes complications and thyroid, and the longitudinal context that makes endocrine reasoning visible.

The trend-data layer

Endocrine decisions are made on trends, not snapshots. A1C of 7.2 today means little without the prior six values. CGM time-in-range only matters with the trajectory. TSH suppressed today on levothyroxine matters in the context of the prior dose changes.

An endocrine-aware AI scribe should:

• Pull last 12 months of A1C, CGM downloads, glucose log, weight, BP, lipid panel, TSH, free T4, free T3 if available
• Surface the trend in the structured note (not just the latest value)
• Flag clinically-relevant changes: A1C trajectory, weight loss/gain, time-in-range improvement
• Tie the trend to today's medication decision — the documentation that supports payer review

GLP-1 / weight-loss medication documentation

GLP-1 receptor agonists (semaglutide, tirzepatide, liraglutide) are the most prior-auth-aggressive class of 2026. Payers require:

• For diabetic indication: A1C history, current/prior diabetes medications, contraindication documentation
• For obesity indication (where covered): BMI ≥ 30 or BMI ≥ 27 with comorbidity, dietary intervention failure documentation, behavioral support component
• For ongoing approval: weight loss target met (typically 5% in 3-6 months), continued dietary engagement
• For escalation: prior dose tolerance, A1C progression, BMI trajectory

A note that says "patient on semaglutide, tolerating well" is not prior-auth-defensive. A note that says "patient on semaglutide 1 mg weekly for diabetes type 2 (E11.65), A1C trended 8.4 → 7.2 → 6.9 over 6 months, no GI intolerance, BMI 32.4 → 29.1, plan to continue at current dose with quarterly A1C" is.

The endocrine-aware system prompt

You are documenting an endocrinology encounter for billing + prior auth defense + longitudinal care.

INPUT:
- Encounter audio transcript
- Patient profile: age, sex, BMI, dx list
- Last 12 months: A1C, CGM (time-in-range / GMI), weight, BP, lipid panel, TSH, free T4, free T3
- Current medication list with doses
- Prior auth status of any controlled / specialty meds

OUTPUT a structured endocrine note:
1. Subjective: chief complaint, interval history (sx, weight change, hypoglycemic events, dietary engagement, exercise)
2. Trend data: present last 4-6 values for each relevant lab / measurement (cite EHR source)
3. Objective: vitals, exam findings, point-of-care labs if any
4. Assessment by problem with ICD-10 v28-specific codes:
   - Diabetes: E11.x with complication subspecificity (E11.42 with peripheral neuropathy, E11.65 with hyperglycemia, etc.)
   - Thyroid: E03.x / E05.x / E07.x with cause and severity
   - Obesity: E66.x with comorbidity coding
   - Lipid: E78.x with statin therapy notation
5. Plan by problem:
   - Med adjustments with rationale tied to trend data
   - For GLP-1 / weight-loss meds: include payer-defensive documentation (BMI, diabetes status, prior failures, weight loss target progress)
   - Lab orders with target intervals
   - Dietary / exercise counseling content (cite transcript — required for some payers)
   - Specialist referrals (CDE, RDN, ophthalmology, podiatry, nephrology) with reason
6. Patient instructions in patient-facing language
7. Follow-up timing tied to clinical question

For each clinical decision, cite the trend or transcript line that supports it. For GLP-1 / specialty meds, structure the note so a payer reviewer finds prior-auth criteria met without searching.

The thyroid case — trend documentation

Thyroid management hinges on TSH trajectory and free T4 over time. A scribe that records "TSH suppressed, plan continue current dose" loses the documentation context. A scribe that records "TSH 0.18 (suppressed), prior values 0.42 → 0.28 → 0.18 over 9 months on levothyroxine 100 mcg, free T4 1.4 (upper normal), patient asymptomatic of hyperthyroidism, plan reduce to 88 mcg with TSH recheck in 8 weeks" produces audit-defensible documentation that supports the dose decision.

CGM data integration

For type 1 and insulin-dependent type 2 patients, CGM data is the primary visit input. The endocrine scribe should:

• Pull the most recent CGM download (Dexcom Clarity, Libre View, etc. via FHIR or vendor API)
• Extract: time-in-range (70-180 mg/dL), time-below-range, time-above-range, GMI, average glucose, glucose variability
• Compare to prior CGM downloads
• Tie to today's insulin / medication decisions

This integration alone justifies the DIY stack for an insulin-pump-and-CGM-heavy endocrine practice — the data is structured, the API is available, and the documentation lift is meaningful.

Vendor matrix — endocrinology AI scribes 2026

For 1-3 endocrinologist independent practices — especially with CGM-heavy panel — the DIY stack with CGM API integration outperforms general scribes meaningfully.

The obesity-medicine angle

Many endocrine practices added obesity medicine to the practice profile in 2024-2025 with the GLP-1 expansion. The documentation rules differ from diabetes:

• BMI documentation at every visit (height + weight + computed BMI)
• Comorbidity documentation justifying obesity treatment (HTN, lipid, OSA, NAFLD, OA, etc.)
• Behavioral component documentation (dietary plan, exercise, behavioral health support)
• Outcome tracking with target percentage weight loss
• Continued benefit documentation for ongoing approval

The endocrine scribe schema can include the obesity-medicine fields conditionally (when obesity is a problem on the list). One prompt, conditional content, full coverage.

BAA and CGM data flow

CGM device companies (Dexcom, Abbott / Libre, Medtronic, Senseonics) offer FHIR-style or vendor APIs for clinical data pull. BAAs are standard across these vendors. The endocrine practice's BAA chain looks like:

Practice + EHR vendor + CGM vendor (Dexcom Clarity, Libre View, etc.) + transcription vendor + LLM vendor.

4-5 BAA documents to manage. Each one is signed once per practice.

When to start

If your endocrine practice manages a high-volume diabetes panel (more than 250 active diabetes patients) or has built up an obesity medicine practice with GLP-1 prior auth volume, the endocrine-aware DIY scribe stack pays for itself in the first month through:

• Faster prior auth turnaround (defensible documentation auto-populated)
• Fewer denials and re-submissions
• Better trend integration in the visit note (less searching the EHR mid-visit)
• Audit-defensible structure for the chronic-condition specialty visits