Clinical Note

Clinical notes do not exist in any documented form. Providers remember what happened during encounters and relay information verbally at shift change. When a consulting physician needs to know what the primary team found, they walk over and ask. The patient's clinical story lives entirely in the heads of whoever was present.

None — AI cannot process, summarize, or learn from clinical encounters because no clinical notes exist in any system.

Providers document encounters inconsistently. One physician dictates detailed notes that get transcribed days later, another writes brief handwritten entries, a third types free-text notes in the EHR with no consistent structure. The quality and completeness of clinical notes depends entirely on who wrote them and how much time they had.

AI could attempt to extract clinical information from the notes that do exist, but inconsistent formats and varying completeness make NLP extraction unreliable — critical findings may be buried in dictation or absent entirely.

Clinical notes follow standard templates in the EHR. Progress notes use SOAP format, admission H&Ps have required sections, and discharge summaries follow a defined structure. All providers document in the same system with the same templates. Note quality is consistent, but the content within sections is still free-text narrative that requires human reading to extract clinical facts.

AI can navigate clinical note sections reliably and extract information from predictable locations (Assessment section for diagnoses, Plan section for orders). NLP can process notes for clinical themes, but precision is limited by free-text ambiguity.

Clinical notes combine structured elements with narrative context. Each note includes coded diagnoses, linked medication references, structured vital sign summaries, and quantified assessment scores alongside the provider's clinical narrative. A query for 'all encounters where sepsis was diagnosed' returns accurate results because diagnoses are coded, not just mentioned in free text.

AI can reliably extract clinical facts from notes using both structured codes and NLP on the narrative sections. Automated coding assistance, clinical summarization, and quality measure abstraction are accurate and actionable.

Clinical notes are schema-driven documents with formal entity relationships. Every clinical assertion links to supporting evidence — a diagnosis links to the lab results and imaging findings that support it, a treatment decision links to the clinical guideline applied, and the provider's reasoning chain is encoded as structured metadata alongside the narrative. AI can ask 'why was this antibiotic chosen?' and trace the reasoning.

AI can perform deep clinical reasoning from notes — understanding not just what was decided but why. Autonomous quality review, coding validation, and clinical documentation improvement are highly accurate because the reasoning structure is explicit.

Clinical notes generate automatically from clinical workflows. Ambient AI captures provider-patient conversations, structures clinical findings from examination data, and assembles comprehensive encounter notes in real-time. The provider reviews and attests to the AI-generated note rather than creating it from scratch. Notes self-update as new information arrives during the encounter — the note is a living document until it is signed.

Can autonomously generate, structure, and maintain clinical documentation from ambient encounter capture. AI produces notes that meet regulatory and billing requirements while preserving clinical nuance and provider reasoning.

Ceiling of the CMC framework for this dimension.