Curated, physician-tested AI workflows for clinical practice, documentation, and research. Validated templates and integration guides built by clinicians for real workflows.
| Workflow | Category | Model / Tool | Time Saved | Validated By | Status |
|---|---|---|---|---|---|
| SOAP Note Drafting — Outpatient FM | Documentation | GPT-4o / Dictation | 18 min/visit | CQ, MD · JR, DO | Active |
| Prior Authorization Letter Generator | Admin | GPT-4 Turbo | 35 min/request | CQ, MD · TL, MD | Active |
| Radiology Report Summarizer | Imaging | Claude 3.5 Sonnet | 8 min/report | CQ, MD | Active |
| Differential Diagnosis Expander | Clinical Reasoning | GPT-4o | 12 min/case | CQ, MD · SA, MD | Active |
| Discharge Summary Automation | Documentation | GPT-4o / EHR API | 22 min/patient | TL, MD · JR, DO | Beta |
| Patient Education Material Generator | Education | Claude 3.5 Sonnet | 20 min/document | CQ, MD | Active |
| Surgical Dictation Structuring | Documentation | Whisper + GPT-4o | 14 min/case | SA, MD · MV, MD | Active |
| IRB Protocol Draft Assistant | Research | GPT-4 Turbo | 3–5 hours/protocol | CQ, MD | In Review |
| Medication Reconciliation Checker | Safety | GPT-4o + Drug DB | 9 min/patient | TL, MD · JR, DO | Active |
| Abstract Screening for Systematic Reviews | Research | Claude 3.5 Sonnet | 60–80% of manual time | CQ, MD | Active |
| System | Integration Type | Compatibility | Guide Version | Notes |
|---|---|---|---|---|
| Epic EHR | SmartText + API | Epic 2024+ | v2.1 | Requires IT approval at most institutions |
| OpenAI API | Direct Integration | GPT-4o, GPT-4 Turbo | v3.0 | HIPAA BAA required |
| Anthropic API | Direct Integration | Claude 3.5 Sonnet, Haiku | v1.4 | Preferred for long document processing |
| Dragon Medical | Dictation + Post-Processing | All versions | v1.8 | Works with existing workflows |
Updated prompt templates to include HPI structuring and assessment/plan separation. New version reduces attending edits by ~40% compared to v2.x. Rollout to members Q2 2025.
Completed validation of abstract screening workflow against two published systematic review datasets. Sensitivity 94.2%, specificity 88.6%. Workflow approved for member use in research pipelines.
Added drug interaction flagging layer using OpenFDA database. Initial pilot shows 12% increase in caught interactions compared to standard EHR alerts. Expanding pilot to 3 additional members.
Added templates for 8 additional commercial payers. Average approval rate for AI-drafted prior auths in pilot: 73% first-pass approval vs. 61% baseline. Templates updated monthly.